Abstract
Background
Psychosocial interventions using social robots aim to enhance the quality of life of older adults, particularly those with dementia. While much research has focused on robot–user interactions, fewer studies have explored how health professionals perceive and use social robots in geriatric care settings.
Objective
This study investigates the perspectives of health professionals on the implementation and integration of social robot interventions (SRIs) in dementia care institutions in France. It aims to identify their perceptions of social robots, perceived benefits, and the organizational and ethical issues associated with their use.
Methods
This qualitative study involved 21 interviews and focus groups with 36 professionals from 13 geriatric institutions. A semi-structured interview guide was used to explore how social robots were introduced, how teams adapted, and what benefits and challenges were reported. The guide was informed by the European health technology assessment approach (HTA Core Model), which encourages analysis across multiple dimensions, including clinical relevance, organizational impact, and ethical considerations. Data were analyzed using thematic synthesis.
Results
Participants described robots being used for emotional support, cognitive stimulation, and recreational activities that encouraged social interaction and improved mood. While robots did not replace core caregiving goals, they offered new ways to engage residents and prompted reflection on professional practices and relationships with residents. Reported benefits included increased resident attention and emotional engagement, opportunities for personalized activities, and renewed motivation among staff. Organizational challenges included limited training, time constraints, and unclear strategies for integrating robots into existing routines. Ethical concerns centered on respecting resident autonomy and consent and avoiding activities that might be perceived as infantilizing.
Conclusion
Social robots hold promise as complementary tools in dementia care. Their successful integration depends on technical feasibility, ethical considerations, institutional support, and sustainable implementation strategies rooted in the principles of person-centered care.
Introduction
The demographic shift toward an aging population has led to a marked increase in chronic and neurodegenerative conditions, notably Alzheimer's disease and related dementias. This trend poses significant challenges for healthcare systems, especially in delivering long-term, person-centered care. 1 In response, nonpharmacological interventions have received increasing attention. 2 Among these, socially assistive technologies, particularly social robots, have emerged as innovative tools for improving the quality of life of older adults.3,4
Social robots are designed to interact with users through verbal, nonverbal, and emotional communication. Their embodiments range from humanoid forms (e.g., NAO) to animal-like designs (e.g., PARO, a robotic seal), with applications across therapeutic, recreational, and cognitive domains. 5 Empirical studies have documented the benefits of social robot interventions (SRIs), including reductions in behavioral and psychological symptoms of dementia, enhanced social engagement, and improved emotional well-being.4,6–10 For instance, interventions involving PARO have been shown to reduce stress, negative emotions, and agitation, while enhancing residents’ social engagement and perceived quality of care.10,11 Additionally, the NAO robot has been used to support cognitive stimulation, 12 improve mood, 13 facilitate physical activity through guided movement exercises,14,15 and promote the implementation of precautionary measures among older adults in geriatric facilities during epidemics. 16
Nevertheless, the integration of SRIs into routine care practices remains limited. Despite increasing interest, relatively little attention has been given in the literature to the processes and conditions that support their use in real-world settings. 17 Hung et al. 11 have emphasized the lack of research on the implementation processes of SRIs in care settings. Several barriers have been identified, including high acquisition and maintenance costs, insufficient staff training, the absence of standardized evaluation tools, ethical concerns (e.g., deception, infantilization), and organizational challenges such as time constraints and workforce instability.11,18,19 These challenges highlight the need for a multidimensional analysis that considers the broader systemic, organizational, and ethical factors shaping the adoption of SRIs.
In this context, the perspectives of health professionals—who interact directly with these technologies and play a key role in their integration into care routines—are especially valuable.20–24 However, few studies have examined these perspectives in depth or across diverse institutional settings, and even fewer have addressed the organizational and ethical conditions that shape real-world use. This lack of implementation-oriented research limits understanding of how SRIs can be sustainably adopted and scaled in dementia care. To address this gap, the present study investigates the implementation of SRIs in geriatric care institutions in France, using the European Health Technology Assessment framework (EUnetHTA Core Model) as a guiding structure.25,26 This framework highlights dimensions (clinical, organizational, ethical) that are central to implementation and are commonly used to inform policy and reimbursement decisions.
The study pursued three objectives: (a) to explore professionals’ perceptions of social robots, including intended purposes and perceived benefits; (b) to identify organizational and ethical challenges; and (c) to analyze clinical, technological, organizational, ethical, and other HTA-related dimensions to understand the introduction and longer-term use of SRIs in dementia care settings and to support future implementation strategies.
Materials and methods
Research design
This study was conducted between September 1, 2021, and June 30, 2022, using a qualitative approach that combined semistructured individual interviews and focus group discussions to explore the perspectives of professionals experienced in implementing SRIs in geriatric care institutions in several regions of France, ensuring diversity in organizational contexts beyond the Paris Region metropolitan area. A qualitative methodology was selected to enable an in-depth examination of individuals’ attitudes, experiences, and perceptions regarding the implementation and use of SRIs. 27
To ensure methodological rigor, the qualitative design was coupled with a content analysis framework and a two-phase coding strategy (inductive followed by deductive), detailed in Section 2.4.
Participants
Purposive sampling was employed to recruit professionals with direct experience in implementing SRIs in geriatric care settings. Invitations containing a study overview and researcher contact information were distributed via email and telephone through national care and research networks. Experience with social robots was an essential inclusion criterion, ensuring that participants could provide informed insights into their use in real-world care contexts. To ensure diversity within the sample, institutions representing a range of geriatric care environments, including long-term care units in university hospitals, and both public and private nursing homes, were approached. For this study, no formal sample size calculation was performed, as we used purposive sampling to gather insights from professionals with relevant experience. The exploratory design, absence of standardized scales, and the novel application of the EUnetHTA framework to qualitative data justified a focus on the depth and diversity of insights rather than statistical representativeness.
Participants included head nurses, nurses, nursing assistants, psychologists, psychomotor therapists, geriatricians, and activity coordinators. Inclusion criteria required that individuals: (a) had direct experience with the implementation of SRIs, whether through planning, coordination, delivery, or evaluation; (b) were actively employed in a geriatric care institution or service; and (c) provided informed, voluntary consent to participate in the study.
Participants were invited to take part in either an individual interview or a focus group. Focus groups were arranged when multiple professionals from the same institution expressed interest in participating. Focus groups did not include hierarchical mixing; each group was composed of participants with similar professional roles (e.g., nurses together, nursing assistants together, psychologists together, or medical doctors together). This organization aimed to facilitate open discussion and reduce the risk of hierarchical bias in participants’ responses.
Internal resource persons assisted with participant identification, which may have introduced a selection bias toward professionals already familiar with or supportive of innovation, although we sought to diversify perspectives across roles and institutions.
Interview guide
A semi-structured guide for interviews and focus groups was developed to explore professionals’ experiences and perspectives (see Supplementary Material N.1). To ensure the content validity, comprehensibility, and contextual relevance of the interview questions, a multi-step process was implemented. First, the initial questions were developed by two researchers, SD (psychologist) and FE (social scientist), based on existing literature related to ethical issues in social robotics. Second, the questions were reviewed by additional experts (MPin and ASR) who assessed their clarity, relevance to clinical practice, and coverage of key thematic areas. Third, a pilot phase involving three test interviews with health professionals (HL, MPic, and AH) was conducted to assess question comprehension and interview flow. Minor revisions were made, including simplifying certain terms, refining probe wording, and adding prompts to capture organizational and ethical aspects more explicitly. The final semistructured interview guide was structured around three core themes: (a) the initial uses of the social robot, including how it was introduced, received, and integrated into care routines within the institution; (b) organizational and resource-related changes associated with its implementation, specifically within the teams or units using the robot; and (c) an open-ended conclusion inviting participants to share reflections on the overall experience and recommendations for improving the integration and future use of SRIs in geriatric care. The guide was also designed to encourage participants to critically reflect on their practical experiences with SRIs, including the ethical dimensions of their use in everyday care. Although the interview guide was primarily developed inductively from literature and expert input, elements of the EUnetHTA framework (e.g., clinical, organizational, ethical, and social aspects) were included during question formulation to cover relevant implementation dimensions. The framework subsequently informed the deductive phase of the analysis (see Section 2.4), where predefined EUnetHTA domains were used to classify and interpret the data.
Procedure
Two members of the research team (FE and SD) conducted the interviews and focus groups across participating institutions. The two researchers had no prior relationship with the participants before the study began. Before taking part, all individuals received detailed information about the study's objectives and procedures and provided written informed consent. Participants were reminded of their right to withdraw from the study at any time without providing justification. Anonymity and confidentiality were ensured, and no individuals other than the participants and the researchers were present during the interviews or focus groups. All interviews were audio-recorded with participants’ consent and anonymized during transcription, with only the participant's profession and institution retained as identifiers. No field notes were taken during the interviews or focus group discussions
Each participant was interviewed once, and transcripts were not returned for feedback or revision. Similar themes consistently emerged across interviews, and the thematic structure remained largely stable after the first ten. However, recruitment continued beyond the point of thematic saturation to ensure representation of professionals across a range of roles within the participating institutions. Audio recordings were deleted following transcription.
Data saturation was assessed across professional groups and institutional contexts. Core descriptive themes relating to the introduction and use of SRIs stabilized after the first ten interviews, primarily among nurses, nursing assistants, and psychologists. Subsequent interviews and focus groups, including those conducted in different institutions, confirmed stability in these themes while adding contextual nuances (e.g., organizational constraints, resource availability). Saturation was therefore considered reached at the thematic level across diverse care settings rather than within a single institution.
Data analysis
Two researchers (MPin and ASR) analyzed data from both individual interviews and focus group discussions using a consistent content analysis approach. Throughout the coding process, the two researchers worked independently but in parallel, and regularly compared and discussed their codes to ensure consistency and reliability. Discrepancies were addressed through collaborative discussion, leading to refinement of the coding scheme where necessary. This process fostered a high degree of agreement and helped establish a shared interpretation of the data. Although no formal inter-rater reliability coefficient (e.g., Cohen's kappa) was calculated, reliability was addressed through iterative comparison and consensus-building between coders. Most discrepancies arose when excerpts could reasonably be assigned to different EUnetHTA domains (e.g., organizational aspects versus ethical considerations). For example, statements regarding lack of time to conduct SRIs were initially coded as organizational constraints by one coder and as ethical tensions by another. These discrepancies were resolved through joint discussion and clarification of coding definitions. Coding decisions and rationales were recorded in analytic memos. The analysis was conducted using Dedoose, 28 a qualitative and mixed-methods data analysis software.
To analyze the interview material, we employed a two-phase coding approach combining both inductive and deductive strategies. In the first phase, we grouped participants’ responses according to the main themes of the semi-structured interview guide, organizing the data around the core areas of discussion. We then conducted an initial round of coding inductively, allowing themes to emerge directly from the raw data. This bottom-up process captured unanticipated categories grounded in participants’ own words. A total of 122 unique codes were generated during this phase. These emergent codes are referred to as First coding (inductive): Emerging codes in Supplementary Material N.2.
In the second phase, we applied a deductive coding strategy based on the EUnetHTA (HTA Core Model, Version 3.0) 25 developed by the European Network of Health Technology Assessment. 26 Using a top-down approach, we systematically classified the data according to predefined domains, topics, and issues from the EUnetHTA framework. The EUnetHTA methodological framework provides a comprehensive approach to evaluating the impact of health technologies across multiple dimensions. While primarily intended to support evidence-based decision-making through the systematic assessment of new technologies, the model is also applicable in research contexts for guiding data collection and analysis, as long as its core principles are respected (Goodman). 29
These categories are labeled as Second coding (deductive): EUnetHTA dimensions in Supplementary Material N.2. Eight of the nine EUnetHTA domains were selected to guide the analysis: Current use of the technology (CUR), Technical characteristics (TEC), Safety (SAF), Clinical effectiveness (EFF), Cost and economic evaluation (ECO), Ethical analysis (ETH), Organizational aspects (ORG), and Patient and social aspects (SOC). Because the HTA framework was applied mainly during the analysis stage and only indirectly informed data collection, certain domains of the HTA model (e.g., clinical effectiveness, cost and economic evaluation) were less represented in the dataset, while organizational, ethical, and patient/social aspects featured more prominently in relation to the interview guide content and participants’ experiential orientation.
Each verbatim excerpt from the interview transcripts was coded using a three-level structure: Level 1: One of the selected EUnetHTA domains; Level 2: A specific theme within that domain; Level 3: A corresponding subquestion from the EUnetHTA Core Model (e.g., Level 1: CUR—Current use of the technology; Level 2: Target population; Level 3: A0007—What is the target population in this assessment?). In the second coding phase, a total of 41 assessment issues from the EUnetHTA Core Model were applied to structure the analysis according to its predefined domains and topics (see Table 2).
The combination of inductive and deductive coding proved useful for analyzing the findings. The inductive phase allowed context-specific insights to emerge directly from participants’ experiences, while the deductive phase, guided by the EUnetHTA constructs, provided a structured lens to interpret these findings within an existing evaluative framework. To illustrate the integration of inductive and deductive coding, an excerpt describing the use of the robot compared to animal-mediated therapies as “It's a positive alternative, because even though we have cats, they can’t visit every resident” was first coded inductively as “differentiation from other interventions” and “target population” while during the deductive phase, the same excerpt was then mapped onto the EUnetHTA domains Features of the technology (TEC) and Target population (CUR). This example demonstrates how descriptive, bottom-up codes generated from participants’ language were subsequently classified within the predefined evaluative domains of the EUnetHTA framework. Additional aligned examples are provided in Supplementary Material N.2.
Ethical considerations
Written informed consent was obtained from all the subjects prior to study initiation. Transcripts were fully anonymized, and confidentiality was maintained throughout the study. Ethical approval was granted by the Non-Interventional Research Ethics Committee of Université Paris Cité (IRB: 00012020-108). The study was conducted in accordance with the principles outlined in the Declaration of Helsinki. In addition, a data protection compliance declaration was submitted to the Data Protection Officer (DPO) of the Assistance Publique–Hôpitaux de Paris (AP-HP).
Finally, in some cases manufacturers or distributors provided training and technical support, raising considerations of commercial influences (e.g., promotional framing, reliance on proprietary materials). Although these relationships did not involve financial incentives, they raised ethical questions about transparency, neutrality, and autonomy in technology selection. These issues are examined in the Results and Discussion sections.
Results
A total of 13 geriatric care facilities located in Île-de-France and other regions of France participated in the study, all of which provided care to patients or residents diagnosed with dementia. A total of 44 health professionals representing a range of roles within geriatric care institutions, including psychologists, managers, nurses, and nursing assistants were invited to participate by email and telephone. Of the 44 approached, eight declined due to time constraints or lack of experience in robotic interventions. The final sampling of 36 professionals included 27 women and nine men. Their age ranged from early 30 to late 60 s and their length of time working with robots was spanning from 2 to 10 years (see Supplementary Material N.3).
In total, 21 interviews were conducted, comprising 10 focus groups and 11 individual interviews. The interviews had an average duration of 71 min, with lengths ranging from 32 to 151 min. Detailed sample characteristics are presented in Table 1. Two types of social robots were mentioned by the participants: PARO and NAO (see Figure 1). One institution experimented with both simultaneously. A description of these robots, including their main features, is provided in Supplementary Material N.4.
Social robots PARO (left) and NAO (right) provided in Supplementary Material N.5.
Details of the composition of the sample of the study.
Overview of EUnetHTA codes applied in the results section.
Qualitative multidimensional analysis using the HTA framework
This section presents a thematic synthesis of the key discussion points raised by professionals concerning the acquisition and use of social robots in geriatric care institutions. The synthesis is supported by selected verbatim excerpts from the interviews. Throughout each subsection, professionals’ insights were systematically linked to the corresponding EUnetHTA codes.
All HTA codes used in the analysis are listed in Table 2. In the textual presentation of the results, when a professional's statement aligned with an HTA assessment element, the relevant domain, topic, and element ID were indicated in parentheses following the excerpt.
Professional perspectives on the target population for SRIs
According to the professionals interviewed, social robots such as PARO and NAO were seldom made available to all patients within a given institution. Instead, the robots were frequently used in specialized care units for patients with Alzheimer's disease (CUR—Target Population A0007; CUR—Utilization A0001). Participants reported that targeted conditions for robot use included cognitive, neurological, behavioral, motor, and relational disorders. Inclusion criteria were generally based on the presence of cognitive impairment, typically assessed using the Mini-Mental State Examination (MMSE), or physical disability, evaluated through Activities of Daily Living (ADL) scores (CUR—Target Population A0002, A0007; CUR—Utilization A0001; ORG—Management G0009). When specified, professionals generally described targeting residents with mild to moderate cognitive impairment (e.g., MMSE scores in the range of 10–20), and excluding those with very advanced dementia or severe behavioral disturbances. “We use robots to support people with cognitive impairments. PARO is particularly suitable for individuals experiencing anxiety or depressive symptoms.” (Psychologist) “NAO is helpful for motivating individuals to perform motor exercises.” (Psychomotor therapist)
Main objectives of SRIs
Professionals identified two primary purposes for using social robots in their institutions: therapeutic applications and recreational activities (CUR—Utilization A0001; TEC—Features of the Technology B0004). These purposes varied depending on the institutional context and implementation strategy. In some cases, participants also reported that the robots were used as part of research initiatives (CUR—Utilization A0012).
Use of social robots for therapeutic purposes
In therapeutic contexts, professionals frequently used the NAO robot to facilitate physical exercise sessions. Both NAO and PARO were also employed to alleviate or prevent stress and to help calm residents during moments of agitation or crisis. As one head nurse explained, “It provides a presence at night; it helps to manage stress.” Participants also described the use of robots to stimulate emotional memory—“[Such as] the animal we loved so much: a dog, a cat … PARO works right away” (Nursing Assistant)—to promote cognitive stimulation—“It fits in with cognitive stimulation approaches around reminiscences” (Psychomotor Therapist)—and to support emotional expression—“The patient will really verbalize affects that he would not verbalize without PARO” (Psychologist) (CUR—Target Condition A0002, A0009; CUR—Utilization A0001, A0009).
Some professionals also emphasized the robot's role in managing psychological or emotional states. One psychologist reported using the robot to help distract residents from pain: “It simply distracts from the pain.” Another explained how the robot could be used to facilitate affective regression in a controlled therapeutic setting: “I use the affective regression that PARO induces, but I know it and I control it, to have access to affects that I don't have during a talk without the robot” (Psychologist) (CUR—Target Condition A0002, A0009; CUR—Utilization A0001, A0009).
According to participants, therapeutic sessions involving social robots were typically conducted on an individual basis and lasted between 15 min and one hour (CUR—Utilization A0011). These interventions were primarily led by psychologists and psychomotor therapists, and less frequently by nurses or nursing assistants (CUR—Utilization G0009; TEC—Features of the Technology B0004; ORG—Management G0009). Some psychologists expressed interest in developing formalized outcome indicators to monitor residents’ progress and to facilitate the use of the robot by other staff members in a more structured therapeutic framework (TEC—Investment and Tools Required to Use the Technology B0010).
Use of social robots for recreational purposes
In recreational settings, professionals reported that PARO and NAO supported social interaction and stimulated conversation among residents. “They have fun, discussing with the robot,” noted a geriatrician, referring to residents with mild cognitive impairment. The robots drew attention and elicited emotional reactions—“It's cute!” (Nursing Assistant)—and were described as reducing feelings of loneliness. As one head nurse explained, adopting the resident's perspective: “It [the robot] needs me, it is all alone; by mirror effect, it does the patient good.” Participants also noted that the robots contributed to diversifying recreational activities: “In this institution, they’re always testing new things; the robot is just one more” (Nursing Assistant) (CUR—Target Conditions A0002, A0009; CUR—Utilization A0001).
Professionals also noted that family members occasionally reacted to the presence of the robot during visits, often expressing curiosity or amusement and, in some cases, mild concern regarding infantilization. While these reactions did not directly shape therapeutic protocols, they sometimes influenced how staff framed the activity or decided whether to introduce the robot in group settings.
According to professionals, these recreational objectives were often pursued by incorporating the robot into existing activities to encourage resident participation. For example, social robots were used during writing or reading sessions to engage residents more actively. In some cases, the robots became central to structured group sessions—such as physical exercise or musical quiz activities—designed around specific recreational goals and intended to enhance resident involvement (CUR—Utilization F0001).
Participants indicated that recreational sessions involving social robots typically lasted between 30 min and 1 h and 15 min, and included groups of four to 10 residents (CUR—Utilization A0011). These activities were most often led by occupational therapists or activity staff, who were responsible for designing and facilitating the sessions (TEC—Features of the Technology B0004; ORG—Management G0009).
Shifting applications of social robots between recreational and therapeutic use
The intended use of the robot—whether therapeutic or recreational—was generally defined prior to its introduction within the institution. According to participants, this orientation was shaped by several factors: initial training provided by robot vendors or distributors; the institution's involvement in experimental projects, reflecting a broader openness to innovation; guidance from in-house professionals identified as key resources; and prior experience with similar interventions, such as Snoezelen rooms, which, like social robots, aim to provide sensory stimulation and emotional support (TEC—Features of the Technology B0004; ORG—Health Delivery Process G0004).
However, the boundary between therapeutic and recreational use was often fluid, reflecting both the evolving needs of residents and the adaptive practices of care teams. In some institutions, professionals who initially used the robot to pursue structured therapeutic goals—such as cognitive stimulation or emotional regulation—gradually integrated it into recreational activities to foster engagement and enhance well-being in a more informal setting. Conversely, others began with recreational applications but shifted toward therapeutic use after observing positive effects on residents’ behavior, mood, or cognitive functioning.
Professionals frequently viewed the therapeutic and recreational uses of social robots as interrelated and mutually reinforcing. As one activity leader explained, “Recreational activities are a pretext for therapy,” while a head nurse reflected, “The nurses and nursing assistants used to say recreation was just playing dominoes, but now they see another dimension.” These interconnections were often reinforced through internal training and informal peer mentoring. For example, activity leaders reported learning from psychologists about the therapeutic value of touch during sessions with geriatric patients involving the robot (ORG—Health Delivery Process G0004).
Even when therapeutic objectives were prioritized, professionals often acknowledged the value of recreational robot use in contributing to broader aspects of residents’ well-being. As one nurse observed, “Therapy is associated with general well-being” (EFF—Health-Related Quality of Life D0012). These transitions between therapeutic and recreational applications were frequently aligned with broader person-centered care strategies, in which flexibility in the use of tools was essential to addressing individual needs and enhancing perceived benefits (CUR—Utilization A0012).
Implementation practices also evolved over time in response to the diverse and changing needs of residents. Group activities, initially the preferred format, were sometimes found to be impractical due to differences in cognitive and physical functioning among participants. In response, more individualized approaches were introduced, with one-on-one sessions becoming increasingly common. As one psychologist explained, “The activity leader goes from room to room in the ward with the robot and proposes short entertaining sessions with it to each voluntary patient” (CUR—Utilization G0009; TEC—Features of the Technology B0004; ORG—Health Delivery Process G0004).
Professionals’ perspectives on the benefits of SRIs
Professionals indicated that most residents responded positively to the robot, whereas a minority refused engagement. No consistent demographic or clinical profile characterized this subgroup based on interview data. As one nursing assistant noted, “Most of the residents really enjoy doing activities with the robot; they find it fun and engaging. But there are a few who just don’t want anything to do with it, they say things like, ‘I don’t want to play with a robot,’ and they prefer more traditional interactions” (EFF—Patient Satisfaction D0017).
Professionals reported a range of perceived benefits associated with SRIs, including improved communication, enhanced attention and responsiveness, reduced behavioral symptoms such as anxiety and agitation, memory stimulation, a general sense of well-being, and the elicitation of positive emotions (CUR—Target Condition A0005; EFF—Morbidity D0005; EFF—Health-Related Quality of Life D0012). Some participants observed that SRIs could encourage socially withdrawn residents to re-engage with staff or fellow residents. The robot frequently served as a conversational trigger, eliciting reactions such as surprise, reminiscence, or even rejection—all of which were considered potentially meaningful within a therapeutic context (CUR—Target Conditions A0009; CUR—Utilization A0001).
Few professionals reported using validated instruments to assess the effectiveness of SRIs in reducing morbidity or improving clinical outcomes. Only one clinician mentioned employing the Neuropsychiatric Inventory (NPI) to evaluate behavioral symptoms before and after the intervention. In most cases, assessments relied on informal or nonstandardized indicators related to quality of life, functional ability, and physical or psychological well-being, and were commonly informed by qualitative observations of residents’ interactions with their environment. Evaluations were typically based on clinical judgment within day-to-day care practices, relying on professionals’ interpretations of observable cues such as physical responsiveness or emotional expression. As one psychologist noted, “We are not into evaluation scores; we do not expect score improvements” (TEC—Investments and Tools Required to Use the Technology B0010).
Organizational practices surrounding the use of SRIs
Knowledge transfer in the context of SRIs: training, practices, and experiential learning
For both PARO and NAO robots, initial training was typically provided by the vendor at the time of purchase (ORG—Health Delivery Process G0003). “The salesman came to do a training session, maybe a week after the robot was introduced in the facility,” recalled a nursing assistant, referring to PARO. Training sessions generally lasted half a day for PARO and a full day for NAO.
Training for the PARO and NAO robots typically consisted of both theoretical and practical components. The theoretical component addressed topics such as emotional engagement, target populations, and professional attitudes toward robot use, while the practical component focused on technical aspects, including cleaning, maintenance, and applied use scenarios. Although professionals generally acknowledged the importance of training for the successful integration of social robots, many found the content insufficient in depth and relevance. Some criticized the limited theoretical input. As one psychologist noted, “There was only technical training which, in my opinion, took ten minutes and a few use cases.” Others pointed to a mismatch between the training materials and the varied educational backgrounds of staff. A psychomotor therapist remarked, “It may be simple for therapists, but not all nursing assistants are comfortable with written content.” Some professionals also found the sessions tedious and insufficiently aligned with their clinical practice. As one head nurse observed, “The session dragged on too long; more time should have been dedicated to hands-on practice to help staff become comfortable with the tool.” These reflections underscore the need for training programs to be better tailored to the diverse roles, competencies, and practical needs of healthcare professionals to ensure relevance and applicability in clinical settings (TEC—Training and Information Needed to Use the Technology B0013; ORG—Health Delivery Process G0003).
Additionally, because training was typically conducted immediately after the robot's acquisition and offered to only a limited subset of staff, many professionals, particularly in institutions experiencing high staff turnover, did not receive it. According to participants, the staff groups most frequently excluded from initial training included nursing assistants, night-shift personnel, and recently hired staff. In several institutions, only psychologists, psychomotor therapists, or activity coordinators received formal training, leaving broader care teams without structured instruction. As one nursing assistant stated, “I don’t know about the training, it was done before I arrived.” A psychologist similarly noted, “Only the psychomotor therapist did the training a year and a half ago.” Limited access to initial training was frequently identified as a barrier to broader adoption. As one head nurse explained, “There's little incentive to use PARO because the necessary knowledge wasn’t effectively transferred to the broader care team.” Professionals emphasized that inadequate dissemination and lack of continuity in training can significantly hinder the integration of social robots into routine care practices (ORG—Health Delivery Process G0003; ORG—Management Problems G0008).
Beyond initial training, successful integration of social robots into care routines often depended on preparatory work conducted prior to their formal introduction. In two institutions, professionals, such as psychologists and psychomotor therapists, had the opportunity to work with robots that had been lent to them before purchase. This preparatory phase, initiated by management in light of the significant financial investment, enabled clinical teams to assess the robot's potential impact on residents, adapt usage scenarios to their specific care context, and critically appraise the claims made by vendors. Notably, this included moderating overly optimistic messaging regarding the robot's capacity to restore cognitive function or replicate authentic human interaction. Such early-stage testing supported the development of realistic expectations, alignment with therapeutic goals, and smoother integration into interdisciplinary care plans (TEC—Investments and Tools Required to Use the Technology B0010).
Professional feedback, covering clinical, organizational, and ethical dimensions, was frequently formalized into internal resources such as summary cards, PowerPoint presentations, or staff training modules. These materials were developed by professionals from a range of disciplines, reflecting diverse perspectives and helping to support shared learning and consistent practices across teams. As one psychologist explained, “Internally, we focus on the real-life challenges. We present case studies based on familiar residents. We don’t need to be generic, we get right to the point and use tailored approaches” (ORG—Health Delivery Process G0003; ECO—Resource Utilization E0001). This emphasis on context-specific, experience-based learning, shaped by input from multiple professional roles, helped ensure that training and implementation strategies remained relevant to everyday clinical practice.
Despite strong initial engagement, several institutions reported a gradual decline in the use of social robots over time. This decrease was frequently attributed to a “novelty effect,” whereby initial enthusiasm waned as the robot became a familiar and routine element within the care environment (ORG—Culture G0010; ORG—Management G0008). In many cases, once the initial curiosity among staff and residents diminished, the robot's integration into regular care activities lost momentum. Professionals noted that without sustained motivation, ongoing training, and structured planning, the robot risked being underutilized or relegated to storage. This pattern highlights the importance of long-term strategies to maintain engagement, including periodic refresher sessions to reinforce staff knowledge and confidence, opportunities to share experiences and best practices, and regular updates on how to adapt the robot's use to evolving care needs and therapeutic goals.
Leadership, institutional positioning, and strategic visions for SRIs
Although the introduction of social robots influenced organizational routines, their integration typically complemented rather than replaced existing care objectives. Key clinical and healthcare staff played a pivotal role during the implementation phase, often emerging as internal champions who helped define the robot's role, deliver peer training, encourage its use, and facilitate communication among team members. These actors either assumed this role voluntarily or were designated by institutional leadership. As one psychomotor therapist explained, “I saw myself as a ‘resource person’ because I had found that solution at the time [how to use the robot more effectively]” (CUR—Utilization G0009; ORG—Structure of Health Care System G0101). Within geriatric services, psychologists often acted as key decision-makers regarding when and how the robot was used.
Administrative managers also played a critical role in initiating or supporting the integration of social robots, particularly when the initiative originated from clinical staff. Their responsibilities included securing acquisition funding, applying for external grants, validating usage protocols, and issuing reminders to encourage consistent use (ORG—Structure of Health Care System G0101). Effective coordination between administrative managers and designated key actors within the clinical team was essential to ensure ongoing follow-up and the establishment of feedback mechanisms regarding the robot's implementation and its effects on care delivery. Many professionals viewed this collaborative dynamic as both supportive and instrumental in promoting the sustainable integration of SRIs into everyday clinical routines (ORG—Health Delivery Process G0012).
In parallel, the financial dimension of acquiring social robots occasionally gave rise to internal tensions. Although most staff members were aware that the robot represented a considerable investment, few knew the exact cost. This lack of transparency sometimes fueled skepticism regarding institutional priorities. As one head nurse observed, “The price of the robot was circulating … people were asking why we spent so much money when we could have invested in staff instead” (ORG—Process-Related Cost G0006; ECO—Utilization E0009).
Some professionals were also tasked with managing access and safeguarding the device due to its high cost. To prevent damage or theft, robots were frequently stored in locked cabinets or staff offices. While these precautions were understandable, they sometimes reduced the robot's accessibility for other team members and contributed to hesitancy around its use. This restricted access also reduced spontaneous use and appears to have led to fewer sessions overall, indirectly limiting residents’ opportunities for engagement with the robot. As one geriatric physician recounted, “I entrusted it to a care assistant. It scared me, I told them, ‘You know it's expensive, be careful,’ so I’m not sure she wanted to use it again” (CUR—Utilization G0009; ECO—Utilization E0009; ORG—Management G0008).
Beyond internal debates over cost and resource allocation, administrative managers often viewed the social robot as a strategic innovation that could enhance the institution's external visibility. It was frequently leveraged as a communication asset in media campaigns, through interviews, videos, and photographs, to showcase the facility's modernity and commitment to forward-thinking care (ETH—Benefit-Harm Balance F0011). While this public exposure occasionally increased staff workload or stress, particularly during demonstrations or special events, it contributed to shaping a positive public image centered on openness to technological innovation and human-centered care. In several cases, the decision to invest in a social robot was supported by prior experience with other nonpharmacological interventions, such as multisensory stimulation or therapeutic environments. In institutional networks, pilot projects involving social robots were sometimes extended to additional sites, reinforcing the robot's role not only as a clinical tool but also as a driver of institutional visibility and innovation (ORG—Management Problems & Opportunities G0008).
Operational demands and resource-related challenges in daily practice
While administrative managers often emphasized the strategic value of social robots, professionals on the ground underscored the practical challenges of their implementation. The use of SRIs requires a substantial time investment, not only for direct interaction with residents, but also for preparatory and follow-up activities. These additional demands impact staff workload and underscore the need for institutional planning to allocate adequate resources and adjust workflows accordingly.
For instance, sessions involving the NAO robot frequently required two staff members: one to engage with the resident and another, typically acting as the “Wizard of Oz,” to operate the robot remotely and coordinate its speech and movements during the session. This configuration demanded coordination, rehearsal, and shared familiarity with the session's content to ensure smooth execution. Facilitators were also responsible for developing scripts and interaction sequences in advance, further increasing preparation time. As one activity leader explained, “At the beginning, it made the facilitators’ work heavier. We had to plan everything so the interaction would feel natural and enjoyable” (ORG—Process-Related Costs D0023; ORG—Health Delivery Process G0004; ECO—Resource Utilization E0001, E0009).
Similarly, the use of PARO introduced additional time-related demands, particularly concerning hygiene and maintenance procedures. The robot required thorough cleaning after each individual session, and this requirement was more strictly enforced during the COVID-19 pandemic. These safety measures, while essential, added to staff workload and required clear institutional protocols to ensure compliance. As one nursing assistant noted, “It takes time to clean it after each use by the residents, and we need to be even more cautious since the COVID-19 period” (SAF—Safety Risk Management C0062; ORG—Health Delivery Process G0001; ECO—Resource Utilization E0001).
The robot as a tool that is difficult to define in healthcare practices
For many professionals, the definition of a social robot took shape progressively through hands-on experience, rather than being clearly defined at the beginning of its use. Their understanding evolved through practice, reflection, and adaptation to specific care contexts. To explain what the robot represented in their clinical environment, professionals often relied on analogies. As one geriatric physician explained, “It can be a tool, a crutch,” suggesting that the robot could serve as a form of complementary support that would assist both patients and staff in their daily routines (ORG—Culture G0010).
To make sense of the robot's role or intended function within care settings, staff often drew comparisons with medications, therapeutic interventions, or familiar everyday objects. The most common analogy was with animal-assisted interventions. PARO, in particular, was frequently associated with pet therapy and regarded as a useful complement—more manageable and easier to control, especially in institutional environments. As one psychomotor therapist observed, “It's a good option, because even though we have cats, they can’t visit every resident. The robot gives them an opportunity to talk about their own animals” (TEC—Features of the Technology B0001, B0002). Other professionals drew parallels with nonpharmacological therapies such as Snoezelen environments, empathy dolls, and music therapy, as well as with other types of social robots or even human caregivers themselves (TEC—Features of the Technology B0001, B0002; ORG—Culture G0010). These analogies further illustrated the multifunctional and evolving nature of the robot in clinical practice, highlighting its potential to support a range of therapeutic interactions and social engagement within institutional care settings.
This ambiguity in defining the robot's role was also reflected in the diverse reactions it elicited among professionals. Acceptance varied across roles and individual attitudes, even within the same institution. Diverging perspectives occasionally gave rise to tensions or discomfort among staff. As one activity leader explained, “If you express that you’re uncomfortable with the robot, people say, ‘What's wrong with you? Are you crazy?’ I know not everyone shares the same view, especially in the research [department], where some people think it's fantastic” (ORG—Culture G0010).
Diverse approaches to introducing and framing social robots in care settings
Professionals described a range of approaches to introducing the robot to residents, reflecting different perspectives on communication strategies and clinical practice. Some adopted a transparent approach, seeking to manage expectations and minimize confusion among older adults. As one psychologist explained, “Personally, I try to avoid creating any confusion, I make it clear that the robot isn’t alive.” Others favored a more interpretive, resident-centered approach, allowing for individual meaning-making. A psychomotor therapist shared, “When a patient asks, ‘Is it real or fake?’ I usually respond with, ‘What do you think?,’ so they can interpret it in their own way.” Trainers and experienced staff generally encouraged this flexible and adaptive strategy, recommending that the robot be introduced in a natural and informal manner, avoiding overly scripted or artificial presentations that might confuse residents (TEC—Training and Information Required B0014; SOC—Patients’ Perspectives H0006; SOC—Communication Aspects H0203; ORG—Communication and Cooperation Between Stakeholders G0007).
However, these recommended practices occasionally led to a sense of professional hierarchy, with some staff members feeling that their experience or contributions were undervalued. One activity leader recalled, “When the psychologists, in charge of the robot, talked to me about NAO, I felt like saying, ‘Hey, I’m not eight years old! I’m your colleague, I’m part of the project!’” (ORG—Management G0008; ORG—Culture G0010).
The “plasticity of use” also extended to how the robots were physically positioned during sessions, with varied practices across institutions. In some settings, professionals emphasized placing the robot at eye level to encourage engagement and reduce perceived distance between the resident and the device. As one nursing assistant explained, “I usually put PARO on a table in front of them (patients) and NAO on the floor in front of them.”
In contrast, other professionals chose to place the NAO robot on a higher table to ensure it was visible and accessible to all participants. However, this setup was not endorsed by all professionals. One activity leader expressed their discomfort: “I saw a session where the robot was high up on a table, like on a pedestal in church, and I thought, ‘What is this? Some kind of satanic ritual?’” (ORG—Culture G0010).
Ethical considerations in the use of social robots
Continuity of care values and ethical principles in SRI integration
Professionals agreed that the implementation of SRIs did not alter their core care objectives. Maintaining meaningful social connections for residents remained a key priority, as one nurse emphasized: “What is very important is the social link between the resident, their family, the caregivers, and their whole ecosystem.” They also underlined the importance of the professional–resident relationship, which remained central to care practices regardless of whether SRIs were involved. As one head nurse explained, “[Whether it's with or without the robot], what really matters to professionals is the relationship they build with the resident, it's at the heart of what we do.”
Professionals emphasized that the implementation of social robots did not compromise their core ethical principles, as the robot's use remained under the supervision and discretion of care staff (ETH—Benefit–Harm Balance F0010). Human empathy was viewed as irreplaceable in caregiving. As one head nurse stated, “Empathy is something very specific to humans.” Concerns about potential dehumanization of care were raised only in relation to unrealistic expectations regarding the robot's capabilities. One geriatric physician illustrated this point: “We can clearly see that after five minutes of interaction [with the robot], a lot of residents lose interest. I really don’t think a robot could ever replace a caregiver” (ETH—Respect for Persons F0008).
Moreover, many professionals believed that thoughtful and appropriate use of the robot was essential to ensure clinical relevance and respect for each resident's individual needs. As one psychomotor therapist explained, “We’ve already thought ahead about what we’re going to do [with the robot], who we’re going to involve, and when.” Others emphasized the value of maintaining a diverse therapeutic toolkit. As one psychologist noted, “Every patient is different … that's why I talk about a therapeutic arsenal. We need many tools that we can try.”
A recurring ethical concern involved how to engage with the beliefs or perceptions of residents with cognitive impairments. This included questions about whether to correct a resident's mistaken belief or to support their current understanding when it brought comfort or stability. As one geriatric physician reflected: “There was a time when we thought we had to correct the beliefs of people with dementia, like telling them their parents had passed away. But now we ask: is it really helpful to challenge that belief?” This perspective echoed broader reflections on how to approach residents’ interpretations of the robot itself, particularly when it was perceived as alive or emotionally responsive (ETH—Autonomy F0005; ETH—Autonomy F0007).
Professionals reported paying close attention to both verbal and nonverbal cues when offering the robot to residents, noting that engagement or refusal could be expressed through gestures, facial expressions, or withdrawal. They described adapting their approach in real time, either encouraging further interaction when the resident appeared comfortable or discontinuing the activity when signs of discomfort or disinterest emerged. In practice, consent was treated as a fluid, ongoing process rather than a single formal act.
From an ethical and institutional perspective, some professionals considered family reactions and the responses of other residents when deciding whether or how to introduce the robot in recreational activities. Concerns raised by family members, particularly around infantilization or authenticity, were occasionally discussed within teams, suggesting that these external perspectives can inform internal norms and usage decisions even if they do not directly alter clinical protocols.
Professional boundaries and ethical tensions in robot-mediated care
The introduction of social robots prompted many professionals to critically reflect on their caregiving values, particularly in relation to the ethics of care. A recurring theme was the tension between the robot's simulated emotional responses and the genuine empathy that professionals regarded as central to their role. As one head nurse observed, “The robot's artificial empathy was unsettling,” highlighting professionals’ strong commitment to the authenticity of their empathic engagement with residents (ETH—Autonomy F0005, F0007).
Concerns about infantilization also emerged. For instance, an activity leader expressed initial hesitation upon encountering the robot, questioning whether its childlike appearance could lead to residents being treated in a patronizing manner: “My first reaction was wondering if we weren’t going to infantilize older adults, because it's a robot [NAO] that looks like a little young man.” This concern was also shared by some families, as one animator noted: “We show it to families during visits; they often react very positively, but some are hesitant, fearing that the robot might infantilize the resident” (ETH—Respect for Persons F0008). Interpretations of residents’ reactions to the robot's playful behaviors and toy-like design varied across professional groups. Some activity leaders and nurses regarded residents’ affectionate or amused responses as potentially infantilizing, whereas psychologists and psychomotor therapists more often understood them as expressions of affectivity or emotional openness. This divergence suggests that ethical impacts and lines of questioning around SRIs may differ according to professional role and scope of practice.
Similarly, the issue of deception was raised, with some professionals questioning whether it was ethical to maintain the illusion of the robot being “alive.” As one psychomotor therapist explained: “If they [residents or patients] raise doubts, that's where I don’t lie to them” (ETH—Respect for Persons F0008).
Professionals also pointed out that misleading communications, such as promotional videos or photographs showing the robot moving autonomously, could generate unrealistic expectations. As one nursing assistant explained, “It looks like it's just walking around by itself! That's what the film shows. But it's not like that at all.” To correct such misconceptions, they emphasized that the robot should not be perceived as a third agent in the care relationship, but rather as a tool that mediates interaction. As one nurse stated, “[The robot] It's just the mediator that makes the connection” (ETH—Benefit–Harm Balance F0003).
Despite these clarifications, some professionals occasionally expressed unease about how the robot might alter their perceived professional role. As one psychologist reflected, “Sometimes I feel like I disappear behind the PARO robot.” Similarly, an activity leader remarked, “We’re standing a bit to the side [of the robot] to make sure everything's going well and that we’re not damaging the machine. I didn’t feel like I was in my role anymore, conducting the activity” (ETH—Respect for Persons F0008).
Importantly, professionals who were hesitant or opposed to using the robot were never forced to do so. Participation remained voluntary, respecting individual comfort levels and professional judgment (ORG—Culture G0010).
Collective ethical reflection and institutional support
Ethical considerations surrounding SRIs were not limited to individual reflections but were also addressed at the collective level within care teams. In some institutions, ethical concerns were actively discussed through internal resources and structured conversations. A geriatric physician mentioned that their team had created a short internal ethics guide to respond to questions raised by staff regarding SRIs, and that these discussions were used to reassure and support professionals who were uncertain about the implications of using the robot. As another head nurse explained: “You have to provide support and debriefing, because it has an impact on their professional identity, you can’t introduce it [the robot] just like that” (ETH—Autonomy F0007).
Beyond individual concerns, many professionals emphasized the value of establishing a shared space for collective ethical reflection. Rather than leaving staff to navigate ethical dilemmas in isolation, some institutions intentionally created forums, whether formal ethics committees or informal discussion sessions, where questions around consent, authenticity, or professional identity could be explored collaboratively. These shared deliberations helped normalize doubts, supported moral reasoning, and strengthened team cohesion. As one psychologist explained, “There really needs to be a shared space for ethical reflection, because each person may experience the robot's introduction differently depending on their sensitivity, experience, and role in the team” (ETH—Autonomy F0007).
Discussion
Purposes of using social robots in geriatric care
Relational mediation and the flexible use of social robots in person-centered care
The purposes of employing social robots in our study were diverse but generally converged on the intention to enhance residents’ quality of life through engagement, stimulation, and recognition of their human dignity. Care teams intentionally used social robots to enrich everyday routines and to engage residents who might otherwise be withdrawn, agitated, or emotionally distant. PARO, for instance, was often introduced in contexts where a calming presence and emotional connection were sought,11,30 while NAO was more frequently used in activities involving cognitive or motor stimulation.16,31 In both cases, the robot was treated as a “third party” in the care relationship, a mediator that facilitated interactions that might not have occurred otherwise. This relational mediation created opportunities for professionals to reconnect with care recipients, not merely as service users, but as individuals with emotional, cognitive, and symbolic dimensions.
Indeed, the integration of social robots such as PARO and NAO into geriatric care environments reflects a shift in care practices, moving beyond a purely biomedical framework toward more holistic, person-centered approaches. Within these settings, older adults are often categorized as “vulnerable,” “frail,” or “dependent.” 32 Such terminology, which primarily defines individuals by their age and impairments, risks reducing them to simplistic labels and overlooking their full identity as persons. As several scholars have argued, this framing tends to promote a narrowly clinical perspective, one that inadequately captures the multidimensional nature of older adults, including their social, emotional, and moral identities.33,34 In contrast, the introduction of social robots seems to offer an opportunity to move beyond the limitations of a purely medical model, toward an approach that acknowledges the full complexity of older adults as individuals. Professionals in our study did not view these devices merely as technological aids, but as symbolic actors that could mediate emotional, social, and relational dimensions of care.
Another notable finding of this study concerns the fluidity between therapeutic and recreational applications of SRIs. While professionals initially differentiated between therapeutic sessions, intended to reduce agitation, stimulate cognition, or support motor activity, and recreational sessions focused on novelty, group interaction, and enjoyment, this distinction often blurred in practice. Recreational activities frequently produced therapeutic effects, and therapeutic interventions were experienced as enjoyable and socially engaging. Moreover, the range of interaction formats (e.g., group settings, one-on-one sessions, and informal encounters) enabled professionals to adjust the robot's use to residents’ fluctuating abilities and preferences. This flexibility supported the delivery of personalized and context-sensitive interventions that addressed both individual and collective well-being. The observed “plasticity of use” underscores that social robots are not confined to fixed functions; rather, their value lies in their capacity to adapt to diverse care situations, resident needs, and professional initiative. These findings align with previous studies highlighting the adaptability of SRIs as a key component of their utility in long-term care settings. 35 This finding also reinforces the relevance of person-centered approaches in dementia care, as articulated in Kitwood's work (1997), 36 which emphasized the importance of recognizing the emotional and relational capacities of individuals with cognitive impairments. This perspective is further supported by recent consensus statements advocating for the personalization of nonpharmacological interventions. 37
Although flexibility was generally viewed as an asset, our findings indicate that this “plasticity of use” can generate both positive and negative effects depending on context. Under favorable conditions (e.g., stable staffing, shared knowledge, supportive leadership), it enabled personalization and creative engagement tailored to residents’ needs. In resource-constrained settings, however, plasticity tended to result in limited standardization, uneven diffusion of know-how, and episodic use restricted to specific professionals. Without shared procedures or training, robot use often depended on individual motivation, creating unequal access among residents and staff. This suggests that flexibility should be balanced with basic organizational structures to support equitable and sustained implementation.
At the resident and family level, plasticity and personalization also moderated expectations of SRIs. Promotional depictions of autonomous or emotionally “intelligent” robots sometimes led to disappointment when performance did not align with these images. A minority of residents refused interaction, noting they “did not want to play with a robot,” indicating that SRIs are not universally acceptable or meaningful. Family reactions ranged from curiosity to concerns about infantilization, prompting staff to adjust how the robot was presented. These dynamics raised ethical questions regarding transparency, informed expectations, and respect for resident preferences, underscoring the need for careful communication when deploying SRIs in person-centered care.
Institutional and symbolic functions of social robots
Beyond their therapeutic and recreational applications, social robots also fulfill important institutional and symbolic roles. Their presence can signal an organizational commitment to innovation, personalized care, and human-centered values. As such, robots become integrated into the facility's identity, serving as both internal tools for reinforcing team cohesion and external symbols used in marketing, public relations, and staff recruitment. In this context, the deployment of social robots extends beyond practical functionality, becoming embedded in institutional discourse and contributing to the construction of professional and organizational legitimacy. 38
Observed benefits and care outcomes of SRIs in geriatric care
Therapeutic engagement and symbolic mediation through social robots
Professionals reported a wide range of perceived benefits associated with the integration of social robots into care practices. These benefits included emotional and cognitive stimulation, enhanced social interaction, and organisational support. Social robots were perceived as tools that could foster interaction and promote connection, particularly among older adults at risk of isolation or limited engagement. Such outcomes highlight the multifunctional nature of these devices when thoughtfully embedded in daily routines.
One of the most frequently cited benefits was emotional engagement. Residents with dementia, including those showing signs of apathy, were observed to respond to social robots in ways that were often difficult to elicit through conventional care approaches. These observations confirm findings from previous studies showing that social robots can enhance emotional expression and interaction among older adults with cognitive impairment. PARO's tactile softness and interactive features, in particular, seemed to foster affective engagement even among individuals with advanced dementia.11,30 Professionals noted that residents who were otherwise passive or minimally responsive began to smile, speak, or display affection when interacting with social robots. In some cases, robots served as mediators of reminiscence, evoking personal memories and facilitating storytelling, particularly when residents associated them with past pets or childhood experiences. From a cognitive and motor perspective, robots like NAO were used to support guided activities aimed at stimulating memory, attention, or coordination, further confirming the potential of SRIs to contribute to multidimensional care goals, as also reported in previous studies.16,31
The nonjudgmental and neutral nature of the robot appears to create a safe space for older adults to express themselves without fear of judgment or embarrassment. This fosters open dialogue and emotional release. Residents were observed to project feelings onto the robot, and this act of projection often served as a medium for communication, not only with the robot itself, but also indirectly with professionals and peers. 39 Such interactions support the therapeutic potential of SRIs by facilitating emotional expression and strengthening interpersonal connections.22–24
Building on this emotional openness, the robot also supported symbolic and cognitive engagement. In particular, it helped sustain continuity in the person's psychic and moral life by evoking memories linked to emotionally significant objects or past experiences. This connection often emerged through physical proximity, such as stroking PARO's soft fur, or emotional resonance, like recalling a beloved pet. In these interactions, the robot was not merely a tool but took on the role of a symbolic stand-in for something or someone emotionally meaningful who was no longer present. This “absent presence,” the sense that the robot evokes the emotional trace of a lost or distant companion, made the past feel vivid and emotionally accessible. Through its tactile and behavioral features, the robot could reawaken memories or attachments, allowing the resident to re-engage with parts of their life history. In doing so, it reinforced the idea that the care recipient is not simply a passive subject of care, but an individual with a rich and ongoing personal narrative. As Sartre (1940) 40 suggested in his theory of imagination, an object can temporarily “become” what it represents in the eyes of the individual. In this context, the robot does not just remind the resident of a pet, it momentarily assumes that emotional role, blurring the line between symbolic representation and lived emotional experience.
Between routine and renewal: how social robots may reshape professional practice
The introduction of social robots into care practices appeared to revitalize the daily routines of healthcare professionals. By expanding the repertoire of therapeutic tools available, these technologies opened up new possibilities for meaningful interaction and creative engagement within settings often marked by repetition and procedural constraints. As Laugier suggests, routinization in care work can give rise to a form of “ordinary opacity,” in which the significance of everyday gestures and interactions becomes obscured by their repetitiveness. 41 In contrast, the integration of social robots seemed to reintroduce a sense of purpose and novelty into daily care routines. Professionals described how the presence of the robot often sparked spontaneous reactions, created moments of joy or curiosity, and stimulated interactions that might not have occurred otherwise. These interactions were not only beneficial for residents but also rewarding for staff, who reported feeling re-engaged and inspired in their caregiving roles.
Moreover, the flexibility of SRIs enabled professionals to adapt their use across various contexts, thus fostering professional creativity and agency. Rather than following rigid protocols, staff could tailor the robot's use to align with residents’ preferences and capacities, which in turn allowed professionals to express their caregiving values more fully. In this way, the integration of social robots contributed not only to resident well-being but also to a renewed sense of purpose and engagement in professionals’ daily work.
Previous studies have also shown that robot-assisted interventions can enhance job satisfaction and help restore a sense of meaning among care professionals.30,42 In our study, professionals described the robot as introducing a new dimension to their interactions with residents, one that was both therapeutic and imaginative. Unlike traditional nonpharmacological tools such as music therapy, sensory rooms, or empathy dolls, the social robot occupies a unique ontological position: it is neither animate like a living being nor inert like a conventional object. This ambiguity, which we interpret as a form of “technological strangeness,” seemed to evoke curiosity and fascination. Some professionals described this experience as a kind of “magic” that enriched the care environment and invited more spontaneous, engaging interactions.
Some professionals acknowledged that the robot's effectiveness can be difficult to explain logically. As one put it, “You have to believe in it,” reflecting how its impact often exceeded initial expectations. Even those who were initially skeptical described being surprised by the emotional engagement it fostered and the new opportunities it created for communication and connection. 43 In this sense, social robots offer more than novelty, they may contribute to renewing the meaning of care itself. This renewal is not merely superficial. As Boudon suggests, shared belief in a tool's effectiveness within a professional culture can shape its impact. 44 When thoughtfully embedded in daily routines, the robot functions as a symbolic mediator, concentrating intention, presence, and empathy in the care relationship. 45 This perspective challenges widespread concerns about the potential for social robots to dehumanize care or erode professional expertise, as raised by scholars like Turkle (2006). 46 Instead, the experiences described by professionals in our study suggest that, when implementation is thoughtful and well-supported, these technologies can reinforce rather than diminish humanistic values. Rather than replacing relational care, social robots may enhance professionals’ sense of purpose, fostering renewed commitment, creativity, and pride in their caregiving roles.
Institutional positioning and the visibility of innovation
From the perspective of institutional managers and administrative staff, the integration of social robots contributed to enhancing the external image and strategic positioning of the facility. These technologies were often presented as visible signs of innovation and personalization in care, frequently featured in communication materials such as promotional videos, interviews or reports. As reflected in the interviews, managers viewed these communication efforts as opportunities to publicly reinforce the facility's modern, human-centered approach.
In several cases, the decision to acquire one or more robots built on prior institutional experience with nonpharmacological technologies, and was supported by leadership committed to investing in gerontechnological innovation. In these contexts, the robot was not only a therapeutic tool but also a strategic asset, aligned with institutional development goals. Based on these findings, one could hypothesize that the adoption of social robots helped certain institutions construct a narrative of modernity, signaling their engagement with quality of care, well-being, and personalized geriatric care. The diversity of available interventions contributed to this image, suggesting a move away from standardized care toward individualized support.
Such framing may also serve practical functions. In a context of growing workforce shortages, visibly innovative practices could enhance the attractiveness of the institution to prospective employees and partners. As suggested by Granjon, modernity can operate as a symbol that makes abstract values, such as care quality, tangible, and publicly legible. 47
Despite these promising observations, our findings indicate that robot-assisted initiatives remain fragile over time. Professionals reported that initial enthusiasm among residents, staff, and leadership often declined as novelty faded and organizational constraints re-emerged, reflecting patterns described in diffusion-of-innovation research in healthcare.
Sustained use was hindered by discontinuous funding, staff turnover, shifting care priorities, and the episodic nature of pilot projects. Robots were frequently acquired through temporary grants or research collaborations rather than recurring budgets, making maintenance, training, and replacement uncertain. Dependency on external funding, combined with leadership changes, evolving care policies, and limited formal procedures for integration into care pathways, interrupted usage trajectories even after early successes.
These dynamics suggest that adoption is not only a matter of technological acceptance but also of organizational sustainability and policy support. Ensuring long-term viability requires planning beyond pilot phases, including stable funding mechanisms, dedicated training structures, and explicit integration into care models rather than symbolic demonstrations of innovation.
Institutional communication strategies can enhance visibility and reinforce innovation narratives, yet they also raise ethical considerations related to privacy and the use of residents’ images for promotional purposes. Filming or photographing residents during innovative practices such as SRI sessions require careful consent procedures, particularly in dementia care settings. Clear institutional guidelines are therefore needed to ensure that communication practices remain respectful, transparent, and consistent with ethical standards of dignity and confidentiality.
Organizational issues
Revealing and reinforcing existing organizational constraints
While professionals did not report that social robots generated new organizational tensions, their introduction often exposed and exacerbated pre-existing ones. These typically involved limited resources, such as time, staffing, and training, and occasionally highlighted hierarchical dynamics among professional roles within institutions.
One of the most frequently cited obstacles was time management. Although robots were generally well received, their use required significant preparation: scripting sessions with NAO, cleaning PARO after use, and coordinating staff schedules. These demands often clashed with the realities of staffing shortages, rotating shifts, and disruptions such as the COVID-19 pandemic. 48 As a result, robot-assisted sessions were sometimes postponed or cancelled, undermining consistency and reducing their impact. Prior studies have similarly noted the additional workload that social robots can introduce in geriatric settings.11,21,49 These difficulties are accentuated by the ongoing shortage of medical professionals willing to work in elder care, a gap that has prompted multiple public policies aimed at improving the sector's attractiveness and perceived value.21,50
Training and knowledge transfer also emerged as persistent organizational challenges. In many cases, only a few designated “resource persons” (such as psychologists or managers) received formal training, resulting in knowledge silos that limited wider staff engagement with the technology. As our findings indicate, this selective training approach restricted the broader care team's ability to integrate social robots into daily routines. High professional turnover further exacerbated the problem: as trained individuals left, incoming staff often received little to no instruction, leading to a gradual erosion of expertise and a decline in robot usage. These dynamics reflect what has been described in the literature as “pilot project syndrome,” wherein promising innovations are introduced with initial enthusiasm but without the necessary long-term infrastructure to ensure sustainability. This reinforces calls from previous research for dedicated resources and institutional support to facilitate ongoing training and capacity-building in the use of social robots.10,11,19,21,49,51,52
Issues of access and equity also emerged during implementation. In several facilities, robots were kept locked away to prevent damage or misuse, and their operation was limited to specific professional roles. While these precautions were often well-intentioned, they inadvertently created disparities between staff members and reduced the opportunities for informal or spontaneous use, interactions that had initially sparked enthusiasm and engagement. As highlighted in our findings, this concentration of expertise and authority around the robot risked reinforcing existing professional hierarchies, rather than fostering inclusive collaboration and shared ownership across the care team.
In summary, the successful introduction of a robotic device in an institution requires not only careful planning but also ongoing monitoring and adaptation over the long term. This ensures that the technology integrates effectively into care routines, aligns with institutional goals, and addresses the evolving needs of both residents and staff.52,53
Difficulties in measuring and validating the outcomes of SRIs
Finally, implementation was rarely accompanied by follow-up mechanisms or structured evaluation. Although qualitative benefits were reported, there was little systematic collection of outcome data, and few institutions used standardized tools to monitor impact. This made it difficult to justify ongoing investment or to adapt practices based on resident responses. Although clinical outcomes were primarily described qualitatively by professionals, future implementations of SRIs could benefit from systematic outcome measurement aligned with the intervention's objectives. For example, cognitive stimulation and behavioral symptoms could be monitored using validated tools such as the Mini-Mental State Examination (MMSE), the Neuropsychiatric Inventory (NPI), or the Cohen–Mansfield Agitation Inventory (CMAI). Emotional well-being and quality of life effects could be assessed using instruments such as the Geriatric Depression Scale (GDS), the Quality of Life in Alzheimer's Disease scale (QoL-AD), or observational affect scales. Collecting such measures would strengthen clinical evaluability and facilitate comparison across settings. As in many healthcare innovations, the lack of outcome data weakened the feedback process necessary for continuous improvement. As Petersson et al. (2022) 54 argue, both quantitative evaluation of clinical and organizational effects and qualitative assessment of healthcare professionals’ and patients’ experiences are essential to ensure meaningful and sustainable implementation. Furthermore, validation and ongoing evaluation of technologies, both before and after their integration into routine care, are needed to demonstrate quality improvements and optimize resource use.
More broadly, the clinical validation of social robots’ impact is challenged by the diversity of care settings, therapeutic goals, and comparator interventions. Standardized assessments often fail to fully capture the personalized, context-sensitive, and multidimensional outcomes that characterize SRIs. As a result, while the benefits may be apparent to professionals, translating them into generalizable evidence remains a significant barrier to broader adoption and policy support. Consequently, the clinical validation of these practices necessitates a more nuanced, multifaceted approach to assessment, one that accounts for the specific context, population, and intended therapeutic goals.3,4,9
Given the HTA framework adopted in this study, future evaluations could more explicitly align mixed-method data collection with specific HTA domains and time points. Quantitative measures such as behavioral symptom scales or time-stamped resource utilization data could inform the Clinical effectiveness (EFF), Organizational (ORG), and Economic (ECO) domains, while qualitative interviews, observations, or video-based analyses could deepen understanding of Ethical (ETH) and Patient/social (SOC) dimensions. Collecting such data at baseline, mid-implementation, and post-implementation would strengthen temporal comparability and allow assessment of both immediate and sustained effects. This structured approach would support more robust HTA-informed evaluations of SRIs in geriatric care.
In the context of cost and efficiency analysis, our findings suggest that several categories of resource use are relevant for future evaluations. These include direct time-related costs (e.g., scripting and operating NAO, cleaning and maintenance for PARO), training and knowledge transfer demands, and organizational coordination activities associated with scheduling and interdisciplinary communication. While our qualitative design did not allow cost quantification, identifying these components provides a preliminary foundation for future economic evaluations of SRIs in geriatric care.
Ethical issues
Ethical reflection emerged as one of the most nuanced dimensions of professionals’ experiences with social robots, revealing how their integration into care required ongoing dialogue to ensure alignment with core care values such as autonomy, relational integrity, and respect for older adults. This process was shaped not only by institutional priorities but also by professionals’ personal values and experiences, highlighting the need for ethical deliberation to be embedded in the design and implementation of SRIs.
Navigating professional identity in relation to SRIs
As previously discussed, one recurring challenge reported by professionals was the difficulty in defining the nature and role of the robot. This ambiguity persisted throughout their experiences, reflecting the ontological indeterminacy of social robots, neither fully animate nor entirely inert. Their symbolic significance and technical complexity often made it challenging for caregivers to clearly situate them within the therapeutic landscape. As Emmanuel (2021) 55 points out, the robot's identity is shaped by a series of implicit design decisions that may not always align with caregivers’ professional values or clinical aims. This misalignment reinforces the need for ongoing ethical reflection and participatory dialogue during the implementation of social robot interventions. 46 Moreover, the robot's ambiguous status, situated between machine and “partner,” can unsettle established norms and expectations within care relationships. 56
These uncertainties are particularly challenging in professions grounded in the ethics of care, where relational presence, empathy, and intuition are essential to practice. The robot, functioning as a third-party entity with preprogrammed behaviors, may inadvertently disrupt the practitioner–care recipient dynamic. While many professionals ultimately came to view the robot as a helpful tool, their initial hesitation often revealed deeper philosophical concerns about its compatibility with human-centered care principles.21,24,56–58
Reaffirming the key role of professionals in care delivery
Recurring lack of clarity regarding the robot's operational role in care delivery, its place within professional relationships, and its strategic purpose within the institution also had direct implications for professional well-being. Some professionals voiced concerns about being replaced or marginalized, particularly in contexts where the robot was prominently featured in media or institutional campaigns. These campaigns often emphasized technological innovation, sometimes overshadowing the emotional labor and relational expertise that caregivers contribute. In facilities already facing staffing shortages or high workloads, this ambiguity could lead to feelings of devaluation or a diminished sense of purpose. While many professionals ultimately came to view the robot as a helpful tool, its unclear positioning occasionally challenged their professional identity and confidence in their caregiving role.11,21
Respondents emphasized the importance of presenting the robot not as an autonomous agent, but as a tool, an instrument of care whose effectiveness depends entirely on the professional's intentions and guidance. As Damour highlights, science fiction-inspired narratives about artificial intelligence often project symbolic meanings onto robots that far exceed their actual capabilities, blurring the line between imagination and reality. 59 These narratives can shape unrealistic expectations and contribute to misunderstandings about what robots can and cannot do in care contexts. In this light, several professionals underlined the importance of grounding social robotics in ethically reflective and pragmatically oriented practices, ensuring that care remains fundamentally human-driven, with technology supporting rather than supplanting professional presence, judgment, and values.
Diverse ethical perspectives and the need for shared frameworks in robot-assisted care
The diversity of professional perspectives on the use and meaning of social robots was also reflected in how caregivers interpreted their ethical implications. Concepts such as regression, infantilization, consent, and autonomy were frequently mobilized but rarely defined in a consistent or shared way across roles or institutions. For example, while some psychologists viewed regression as a beneficial affective state that allowed residents to access deep emotional expression, others considered it a potential threat to dignity and personhood. These divergent interpretations reflect deeper disciplinary ethical orientations. For example, psychologists and psychomotor therapists often draw on psychodynamic or person-centered traditions in which temporary regression or childlike engagement may be framed as pathways to affective expression, therapeutic insight, or emotional resonance. In contrast, nurses and activity leaders frequently operate within ethical frameworks that prioritize dignity, autonomy, and the avoidance of behaviors perceived as infantilizing. These disciplinary cultures shape how the same resident behaviors are morally evaluated, helping explain why regression can be viewed as therapeutic in one profession and degrading in another. These differing interpretations illustrate not a confusion or difficulty in positioning, but rather the coexistence of multiple ethical frameworks and clinical paradigms within care teams. This underscores the need for shared spaces of ethical reflection and a common language that enables professionals to collectively explore and navigate the complex moral dimensions of robot-assisted care.19,60
A related dilemma concerned the risk of infantilization. While some professionals embraced residents’ affectionate or playful interactions with robots like PARO, others expressed unease, fearing these responses might reinforce reductive or ageist portrayals of older adults. The robot's design, vocalizations, and scripted behavior often played into these concerns. To address this, some professionals sought to adapt their use of the robot—ensuring it was positioned at eye level, using respectful and age-appropriate language, and supporting residents’ interpretations without imposing predefined meanings.
Respect for residents’ rights and autonomy was another core ethical consideration. Even when cognitive impairments were present, professionals emphasized that older adults retain the right to consent to care practices, including interactions with social robots. However, this raised important questions: How should consent be obtained in practice? Can spontaneous emotional responses be interpreted as valid indicators of consent? Engagement with the robot was often immediate, nonverbal, and fluid—challenging conventional notions of informed consent.
To address these complexities, professionals in our study described conceptualizing consent as a dynamic and context-dependent process. Rather than relying solely on formal procedures, they attended carefully to both verbal and nonverbal cues, adapting their approach in real time. Some chose to clarify explicitly that the robot was not alive, prioritizing transparency and cognitive clarity. Others, however, allowed residents to interpret the robot in their own way—respecting their imagination and personal understanding of the experience, especially when such interpretations appeared comforting or emotionally beneficial.
Moreover, professionals stressed the importance of reassessing consent at each encounter, acknowledging that a resident's capacity or willingness to engage with the robot may vary from day to day. This flexible, relational approach aligns with recommendations from previous literature that call for a more nuanced understanding of consent in dementia care contexts.23,24,34,56,57,61
These observed practices align with established frameworks in dementia care that conceptualize consent as dynamic rather than static. For example, Dewing's “process consent” model emphasizes the ongoing monitoring of more-than-verbal indicators of willingness, comfort, or refusal, while assent/dissent frameworks recognize the ethical validity of nonverbal participation or withdrawal, particularly in advanced cognitive impairment. 62 Connecting our findings to these models highlights that although participants did not explicitly use such terminology, they nevertheless implemented consent as an iterative and relational process.
Professionals also raised concerns about financial accessibility. The cost of social robots was frequently described as prohibitively high and not covered by public health insurance, limiting equitable access across care institutions. This raised ethical questions about fairness and justice: how can promising innovations be made available to all older adults, regardless of an institution's financial capacity? Professionals expressed concern that, without adequate funding mechanisms, these technologies risk exacerbating existing disparities rather than reducing them. Similar concerns have been echoed in the literature, where cost is consistently cited as a major barrier to adoption.21,23,24,51,63–65 Some authors have highlighted the importance of assessing the costs of social robots in relation to the benefits they may bring when introduced in geriatric settings. 66 In response, researchers have proposed the use of lower-cost alternatives 63 to improve accessibility in under-resourced settings. Nonetheless, in the absence of public reimbursement pathways or long-term institutional investment, professionals noted that the use of social robots remains largely contingent on local budgets and temporary external funding opportunities.
Ethical evaluations were often grounded in immediate impressions and professional intuition rather than in formal ethical frameworks. While such an experiential approach aligns with the relational and context-sensitive nature of care, it may risk overlooking broader structural issues or inconsistencies between institutions. To support more consistent and reflective practices, professionals emphasized the importance of creating dedicated spaces for ethical deliberation—such as training sessions, interdisciplinary discussion groups, or written guidelines. These resources could help formalize ethical reflection and ensure it is sustained as an integral part of the implementation process.21,23,24
From implicit ethics to shared reflection in the implementation of SRIs
The integration of SRIs in geriatric care prompted professionals to reflect more explicitly on the ethical dimensions of their practice, particularly around social dignity.33,34,38 For many, the robot served as a catalyst that surfaced or reframed existing moral evaluation criteria, often previously implicit, around what constitutes appropriate and respectful care. These criteria drew from three principal sources: the professional's caregiving intentions, intuitive assessments of contextual appropriateness, and real-time observation of residents’ emotional or behavioral responses. This form of ethical reflection was grounded less in rigid principles and more in lived experience and moral attentiveness. It corresponds to what Laitinen et al. 34 describe as a “recognition ethics,” where moral value arises not only from fulfilling care tasks, but from acknowledging the personhood of the resident through relational engagement.
Although much of this ethical reasoning appeared intuitive, it was frequently supported and enriched through structured team discussions. Many institutions encouraged such reflection by organizing feedback meetings, drafting shared guidelines, or creating informal discussion spaces. These deliberative practices helped align ethical standards across teams while preserving flexibility to adapt to individual resident needs. This approach resonates with the recommendations of Koh et al.,23,24 who advocate for institutionally supported, team-based ethical reflection in the face of emerging care technologies.
Rather than treating ethical tension as a problem to be resolved, professionals engaged with it as a generative space, examining the evolving meanings of concepts like “regression,” “autonomy,” and “consent.” As discussed by Frennert et al., 21 the arrival of social robots in care environments frequently prompts caregivers to rearticulate their understanding of what it means to care and to consider whether their interventions continue to support or compromise the integrity of the patient. This “value work” strengthens ethical reflexivity and affirms the importance of collective reflection as part of caregiving practice. 38
In this light, the introduction of social robots did not dilute the ethical grounding of care but rather reinforced it, provided that institutions fostered shared spaces for reflection, dialogue, and meaning-making. The generally positive attitudes observed among professionals in this study are consistent with findings by Chen et al. (2020) and Parviainen et al. (2019), which suggest that increased familiarity with social robots tends to support more accepting and constructive engagement with them.20,67 Enhancing ethical literacy is therefore essential to the responsible integration of SRIs.19,20,38 As one concrete response, Papadopoulos et al. 60 developed a dedicated training program to strengthen healthcare professionals’ capacity to navigate ethical issues raised by SRIs, an initiative that was well received by participants and offers a model for institutional investment in this area.
While our ethical analysis draws primarily on Western and European frameworks rooted in relational autonomy, dignity, and professional ethics, perceptions of social robots and their moral significance vary across cultural contexts. In East Asian settings, for example, anthropomorphic technologies may be more readily integrated into daily life due to different socio-cultural attitudes toward robots, aging, and human–machine interaction, potentially reshaping ethical concerns and expectations. Incorporating culturally diverse perspectives would therefore enrich comparative studies and clarify how notions of autonomy, infantilization, or emotional authenticity may shift across societies. These ethical considerations operate across clinical, institutional, cultural, and policy levels.
At the policy level, ethical governance is increasingly embedded in European digital health strategies that link procurement, certification, and reimbursement to compliance with ethical and regulatory standards (e.g., GDPR, medical device regulation, AI ethics guidelines). However, for socially assistive robots, explicit accreditation or reimbursement pathways incorporating ethical criteria remain limited. Clarifying these interfaces could support more responsible diffusion by aligning innovation incentives with ethical oversight and patient rights.
Contributions and limitations of the study
The qualitative approach adopted in this study enabled the collection of rich, in-depth perspectives from a diverse group of professionals involved in the implementation of SRIs in geriatric care settings. The participating institutions were varied in size, structure, and level of robot integration, which contributed to capturing a broad spectrum of experiences. Thematic analysis allowed for the identification of complex, interrelated processes associated with the acquisition and use of social robots—offering a more integrated understanding of the challenges, facilitators, and ethical considerations at play, rather than a fragmented or overly deterministic account.
A distinctive aspect of this study was its use of the EUnetHTA Core Model,25,26 which provided a comprehensive and multidimensional framework for organizing data across clinical, organizational, ethical, and economic domains. This HTA-based perspective offered valuable insights into how professionals engage with SRIs not only as technological tools but as interventions embedded within specific institutional and ethical contexts.
Several methodological limitations should be acknowledged. First, the HTA framework was applied unevenly across domains: because the interview guide focused more on organizational, ethical, and experiential aspects of implementation—and professionals emphasized these dimensions—domains such as clinical effectiveness, technical features, and cost/economic evaluation were less extensively documented. Second, although purposive sampling was used, recruitment relied on institutional contacts who recommended professionals identified as “resource persons,” potentially introducing selection bias by favoring more engaged or supportive staff. Third, while thematic saturation was monitored, no formal sample size calculation was conducted, which may affect the transferability of the findings to other care contexts. Fourth, field notes were not taken during interviews and focus groups, limiting the capture of contextual elements such as group dynamics and nonverbal cues; although audio recordings preserved verbal content, the absence of observational data may have reduced dataset richness. Fifth, participant feedback on findings was not sought due to practical challenges in recontacting geriatric care professionals and concerns about participant burden, limiting opportunities for participants to clarify the researchers’ interpretations. Sixth, although professionals noted that a minority of residents declined to engage with SRIs, the data did not allow further characterization of this group, restricting understanding of who may be less likely to benefit. Finally, although data were collected across multiple regions in France, organizational and financing models vary internationally, which may limit the transferability of certain findings.
Conclusion
Interviews and focus groups with healthcare professionals highlight the therapeutic potential of social robots within geriatric care settings. These devices offer flexible, personalized interventions for residents with complex, chronic, and neurocognitive disorders, distinguishing them from standardized pharmaceutical approaches. Their adaptive features and symbolic functions enable professionals to tailor interactions to residents’ emotional states, capacities, and preferences.
A key finding from this study is that social robots operate not only as therapeutic tools but also as mediators of emotional expression, memory, and social engagement. By fostering reminiscence and facilitating communication, they contribute to a more holistic, person-centered approach to care. At the same time, their ambiguous status, between object and social agent, elicits ethical reflection on dignity, authenticity, emotional reciprocity, and consent.
This work identifies several facilitators of SRI integration: stimulation of ethical reflection among staff, support for individualized care practices, reinforcement of an institution's innovative identity, and alignment with broader shifts toward nonpharmacological and autonomy-supportive interventions. Persistent challenges also emerged, including uncertainties about the robot's role, concerns over professional identity, organizational constraints linked to staffing and training, inequalities in access, and the absence of structured funding or reimbursement at the policy level. These findings underscore that the impact of SRIs depends less on the robot alone than on the ethical, organizational, and regulatory frameworks in which it is embedded.
Practical recommendations emerge for key stakeholder groups:
For healthcare administrators: develop recurrent funding and training strategies, integrate robots into existing care pathways rather than isolated pilot initiatives, and ensure equitable access across units and staff categories. For policymakers and HTA bodies: establish clearer evaluation pathways for socially assistive robots, combining clinical, organizational, ethical, and economic indicators, and explore reimbursement models that reflect their multidimensional value. For training and education program designers: incorporate robot-mediated care, dementia ethics, and interdisciplinary collaboration into professional curricula in gerontology, nursing, occupational therapy, and psychology. For researchers and innovation partners: promote mixed-methods evaluation frameworks that pair qualitative insights with standardized outcomes to improve comparability across settings and support evidence-informed policy.
These insights suggest that social robots can enhance quality of care when implemented responsibly, but their long-term contribution depends on sustained organizational support, ethical literacy, and policy alignment to ensure consistency with the core values of dignity, equity, and relational caregiving.
Supplemental Material
sj-docx-1-dhj-10.1177_20552076261428151 - Supplemental material for Health professionals’ perspectives on the implementation of social robots interventions for dementia care in France: An interview and focus group study
Supplemental material, sj-docx-1-dhj-10.1177_20552076261428151 for Health professionals’ perspectives on the implementation of social robots interventions for dementia care in France: An interview and focus group study by Maribel Pino, Faustine Emmanuel, Hermine Lenoir, Matthieu Piccoli, Abdelhakim Hacil, Sebastien Dacunha and Anne-Sophie Rigaud in DIGITAL HEALTH
Supplemental Material
sj-docx-2-dhj-10.1177_20552076261428151 - Supplemental material for Health professionals’ perspectives on the implementation of social robots interventions for dementia care in France: An interview and focus group study
Supplemental material, sj-docx-2-dhj-10.1177_20552076261428151 for Health professionals’ perspectives on the implementation of social robots interventions for dementia care in France: An interview and focus group study by Maribel Pino, Faustine Emmanuel, Hermine Lenoir, Matthieu Piccoli, Abdelhakim Hacil, Sebastien Dacunha and Anne-Sophie Rigaud in DIGITAL HEALTH
Supplemental Material
sj-docx-3-dhj-10.1177_20552076261428151 - Supplemental material for Health professionals’ perspectives on the implementation of social robots interventions for dementia care in France: An interview and focus group study
Supplemental material, sj-docx-3-dhj-10.1177_20552076261428151 for Health professionals’ perspectives on the implementation of social robots interventions for dementia care in France: An interview and focus group study by Maribel Pino, Faustine Emmanuel, Hermine Lenoir, Matthieu Piccoli, Abdelhakim Hacil, Sebastien Dacunha and Anne-Sophie Rigaud in DIGITAL HEALTH
Supplemental Material
sj-docx-4-dhj-10.1177_20552076261428151 - Supplemental material for Health professionals’ perspectives on the implementation of social robots interventions for dementia care in France: An interview and focus group study
Supplemental material, sj-docx-4-dhj-10.1177_20552076261428151 for Health professionals’ perspectives on the implementation of social robots interventions for dementia care in France: An interview and focus group study by Maribel Pino, Faustine Emmanuel, Hermine Lenoir, Matthieu Piccoli, Abdelhakim Hacil, Sebastien Dacunha and Anne-Sophie Rigaud in DIGITAL HEALTH
Supplemental Material
sj-docx-5-dhj-10.1177_20552076261428151 - Supplemental material for Health professionals’ perspectives on the implementation of social robots interventions for dementia care in France: An interview and focus group study
Supplemental material, sj-docx-5-dhj-10.1177_20552076261428151 for Health professionals’ perspectives on the implementation of social robots interventions for dementia care in France: An interview and focus group study by Maribel Pino, Faustine Emmanuel, Hermine Lenoir, Matthieu Piccoli, Abdelhakim Hacil, Sebastien Dacunha and Anne-Sophie Rigaud in DIGITAL HEALTH
Supplemental Material
sj-docx-6-dhj-10.1177_20552076261428151 - Supplemental material for Health professionals’ perspectives on the implementation of social robots interventions for dementia care in France: An interview and focus group study
Supplemental material, sj-docx-6-dhj-10.1177_20552076261428151 for Health professionals’ perspectives on the implementation of social robots interventions for dementia care in France: An interview and focus group study by Maribel Pino, Faustine Emmanuel, Hermine Lenoir, Matthieu Piccoli, Abdelhakim Hacil, Sebastien Dacunha and Anne-Sophie Rigaud in DIGITAL HEALTH
Footnotes
Acknowledgments
We thank our partners in the Rosie project: Gérontopôle d’Ile-de-France, la Chaire Hospinnomics (Paris School of Economics/AP-HP) and l’EA 2694 “Santé Publique: Epidémiologie et qualité des soins” of l’Université de Lille.
Ethical considerations
The study was conducted according to the guidelines of the Declaration of Helsinki, and approved by the Non-Interventional Research Ethics Committee of Université Paris Cité (No. IRB: 00012020-108). In addition, a data protection compliance declaration was submitted to the Data Protection Officer (DPO) of the Assistance Publique—Hôpitaux de Paris (AP-HP).
Consent to participate
Written informed consent was obtained from all participants involved in the study.
Consent for publication
Not applicable.
Author contributions
Conceptualization: MPin and ASR; data curation: FE and SD; formal analysis: FE and SD; funding acquisition: MPin and ASR; investigation: FE, SD, HL, MPic, and AH; methodology: MPin, ASR, FE, and SD; project administration: MPin, ASR, and SD; resources: ASR, HL, MPic, and AH; supervision: MPin and ASR; validation: MPin and ASR; writing–original draft preparation: FE, MPin, and ASR; and writing–review and editing: ASR, SD, MPin, MPic, HL, and AH. All authors have read and agreed to the published version of the manuscript.
Funding
This research was funded by Caisse Nationale de Solidarité pour l’Autonomie (CNSA) and Malakoff Humanis (Grant No. 2017/2017-225/02).
Declaration of conflicting interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Data availability
Study data is available upon request.
Supplemental material
Supplemental material for this article is available online.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
