Optimizing nursing clinical decision support systems: Insights from nurses’ experiences

Abstract

Purpose

This study explores nurses’ experiences with a Nursing Clinical Decision Support System (Nu-CDSS) two years post-implementation and identifies user-centered areas for system optimization.

Methods

A descriptive qualitative study was conducted in a tertiary acute care hospital in Shanghai. A select, stratified group of 15 Chinese-speaking clinical nurses and nurse managers across five wards participated in in-depth semi-structured interviews. Data were transcribed verbatim and analyzed thematically with findings organized using the 5R framework.

Results

Seven themes and 14 subthemes were identified. Optimization requirements encompassed: 1) Right Information: enhancing clinical relevance and patient-specific adaptability of diagnoses and interventions; 2) Right People: optimizing role-based usability; 3) Right Formats: integrating clinical records and embedding notifications directly into primary workflows; 4) Right Channels: improving platform interoperability and seamless mobile-desktop data synchronization; 5) Right Time: minimizing system latency and optimizing reminder timing and lifecycles. Furthermore, participants emphasized the need for 6) robust technical and network support, including adaptive inference engines, and 7) standardized nursing protocols.

Conclusion

Optimizing Nu-CDSS in acute care requires improvements in the knowledge base, interface integration, platform synchronization, reminder timing, and technical support. Future iterations should incorporate AI-enhanced, context-sensitive decision support while preserving nurses’ clinical judgment.

1. Introduction

As nursing informatics continues to evolve alongside rapid healthcare digitalization, the integration of artificial intelligence (AI) has significantly expanded the capabilities of Nursing Clinical Decision Support Systems (Nu-CDSS) to optimize staffing and patient care.^1,2 Comprising a knowledge base, an inference engine, and a human-computer interface, ³ these systems emulate human decision-making by matching patient data with pertinent clinical information. ⁴ Consequently, Nu-CDSS directly assists in formulating nursing diagnoses, care plans, interventions, and outcome evaluations,^5–7 ultimately optimizing clinical workflows, standardizing practices, and reducing the operational burden on nurses. ⁸

However, the implementation and long-term maintenance of Nu-CDSS require substantial financial resources. The expenses associated with adoption, maintenance, and support are significantly high.^9,10 Moreover, aligning these systems with the complex workflow of healthcare organizations presents significant challenges. ¹¹ Studies indicate that the current performance of Nu-CDSS does not fully meet clinical needs, highlighting the necessity for continuous improvement. ¹²

Recognizing these challenges, healthcare institutions worldwide have increasingly implemented Nu-CDSS to support clinical decision-making and enhance nursing practice. ¹³ Despite their potential to enhance efficiency and patient outcomes, Nu-CDSS alignment with nurses’ clinical expectations remains uncertain. Importantly, the successful integration of Nu-CDSS is highly context-dependent. Competencies and technological readiness vary considerably depending on the clinical setting. Acute care nurses exhibit distinct technical competencies and operational demands compared to those in general ward or long-term care. Furthermore, variations in healthcare infrastructure, clinical workflows, and technological adoption across different regions further underscore the need for a comprehensive evaluation of Nu-CDSS implementation and optimization strategies. ¹⁴

To inform context-sensitive optimization, this small qualitative study explored the experiences of a select, stratified group of Chinese-speaking nurses using the Nu-CDSS in an acute care setting. Guided by the 5R framework, we asked: (1) how nurses experienced integrating the Nu-CDSS into daily workflows; (2) where current system functions diverged from clinical needs across the 5R dimensions; and (3) what user-centered improvements were needed to enhance usability and clinical relevance. By investigating nurses’ perspectives, this study seeks to provide insights for refining Nu-CDSS and future AI-enhanced frameworks to better support clinical decision-making in nursing practice.

2. Methods

2.1. Study design

This study used a descriptive qualitative approach, guided by the 5R framework. The study adhered to the Consolidated Criteria for Reporting Qualitative Research (COREQ) checklist to ensure rigor in study design and reporting ¹⁵ (Supplemental File).

2.2. Conceptual framework

The 5R theory proposed by Osheroff et al. ¹⁶ outlines key principles for designing clinical decision support systems. It emphasizes ensuring the right information reaches the right people through the right channels in the right formats at the right times to improve healthcare processes and outcomes. As no validated scale exists for optimizing Nu-CDSS needs, the 5R theory provided a structured framework for data collection and analysis, and served as a deductive lens for mapping qualitative codes into actionable system optimization categories.

2.3. Study setting and participants

The study was conducted in a 4,000-bed tertiary acute clinical care teaching hospital. Participants were recruited from five wards that had used the Nu-CDSS since its rollout, using purposive sampling with a maximum variation strategy to ensure diversity across ward type and clinical role, including clinical nurses and nurse managers.

Participants demonstrated high baseline technological literacy, routinely utilizing electronic health records and mobile clinical devices. Prior to data collection, all had completed mandatory, standardized training covering the Nu-CDSS interface, data input, and alert management. Inclusion criteria required informed consent and at least one year of Nu-CDSS experience for clinical nurses or managers. Nurses on extended leave or lacking sufficient system experience were excluded.

2.4. Data collection

2.4.1. Semi-structured interview

From December 2023 to April 2024, 15 in-depth face-to-face interviews were conducted by the first two female authors, both trained in qualitative research and interviewing; the first author was a PhD student. No pre-existing supervisory relationships existed. Interviews were conducted in private hospital rooms with no one else present. Participants were informed that the interviewers were conducting the study to explore nurses’ experiences with the Nu-CDSS and identify optimization needs. The interview guide (Supplemental File) was developed deductively from the 5R framework ¹⁶ and the research team’s clinical expertise. During interviews, participants were shown representative Nu-CDSS interface artifacts, including the intervention scheduling page and reminder interface, to facilitate context-anchored discussion. Interviews lasted 34–56 minutes and were audio-recorded with consent.

2.4.2. Transcription and translation

All interviews were conducted in Chinese and transcribed verbatim, explicitly capturing non-verbal cues to preserve contextual nuances. The initial transcripts were subsequently returned to participants for verification and modification via member checking to ensure accuracy.

Verified transcripts were translated into English by a qualified bilingual translator. To guarantee conceptual equivalence, an independent translator performed back-translation, resolving any discrepancies. Translated themes and illustrative quotations were reviewed by two independent bilingual reviewers, including one native Chinese speaker and one native English speaker, to improve semantic fidelity across languages.

2.5. Data analysis

Data collection and analysis were conducted concurrently using Colaizzi’s seven-step method (1978). Two researchers independently open-coded the first three transcripts line by line and then compared coding decisions to develop a preliminary codebook. The codebook was iteratively refined through repeated discussion during analysis of subsequent transcripts. Candidate themes were reviewed against the full dataset to ensure internal coherence and fidelity to participants’ accounts. After inductive theme generation, the 5R framework was applied as an interpretive matrix to organize themes into actionable system dimensions rather than to predetermine coding categories. To ensure methodological rigor and interpretative depth, two independent researchers conducted the analysis, with discrepancies resolved through discussion with a third senior researcher. ¹⁷

2.6. Ethical considerations

This study received approval from the Ethics Review Committee (Approval No.: B2023-355R). Written informed consent was obtained from all participants. Participants were assured of the study’s voluntary nature and their right to withdraw without consequence. Confidentiality measures were strictly followed. Any patient data referenced during interviews (e.g., vital signs) came from routine, clinically indicated care rather than research-specific collection, and all such discussions were fully anonymized to address privacy concerns.

2.7. Rigor and trustworthiness

Rigor was established in line with qualitative guidance. ¹⁷ Credibility was supported through reflexivity and member checking of the Chinese transcripts. Dependability was enhanced through an audit trail of data collection and analysis. Confirmability was strengthened by independent coding, third-researcher adjudication, and bilingual translation verification. Transferability was supported by detailed reporting of the setting, participants, and sampling strategy. The team comprised five nurses and one undergraduate student (one male, five females) with expertise in qualitative research, nursing informatics, and Nu-CDSS, and reflexivity was used throughout data collection and analysis to minimize the influence of prior assumptions.

3. Results

3.1. Participants’ characteristics

Fifteen nurses participated (mean age: 38.00 ± 7.62 years; clinical experience: 17.00 ± 9.27 years). Data saturation was achieved as no new themes emerged during the final two interviews, precluding the need for repeat interviews. Detailed characteristics of clinical nurses (CN) and nurse managers (NM) are presented in Table 1.

Table 1.

The participants’ characteristics.

Participant ID	Gender	Age	Role	Educational background	Years of working experience	Duration of using Nu-CDSS (months)
1	Women	40	Clinical Nurse	Bachelor’s degree	19	53
2	Women	53	Nurse Manager	Bachelor’s degree	35	53
3	Women	35	Clinical Nurse	Bachelor’s degree	19	53
4	Women	35	Clinical Nurse	Bachelor’s degree	8	53
5	Women	44	Nurse Manager	Bachelor’s degree	24	53
6	Women	35	Clinical Nurse	Bachelor’s degree	10	53
7	Women	41	Clinical Nurse	Bachelor’s degree	22	53
8	Women	30	Clinical Nurse	Bachelor’s degree	10	53
9	Women	26	Clinical Nurse	Bachelor’s degree	3	28
10	Women	24	Clinical Nurse	Bachelor’s degree	2	22
11	Women	44	Nurse Manager	Bachelor’s degree	25	53
12	Women	41	Nurse Manager	Bachelor’s degree	21	53
13	Women	45	Nurse Manager	Bachelor’s degree	25	53
14	Women	41	Clinical Nurse	Bachelor’s degree	21	53
15	Women	36	Nurse Manager	Master’s degree	11	53

3.2. Overview of themes

Data analysis yielded seven primary themes and 14 subthemes (Table 2). Guided by the 5R framework, the findings highlighted the need for more clinically relevant and patient-specific Right Information, improved role-based usability for the Right People, more integrated Right Formats, more seamless Right Channels, and more timely reminders at the Right Time. Beyond the 5R dimensions, participants also emphasized the importance of technical support, network stability, and standardized nursing protocols.

Table 2.

Summary of themes and subthemes.

Themes	Subthemes
Theme 1: Experiences and Optimization Needs Regarding “Right Information”	Subtheme 1: The Need for Nursing Diagnoses to Align More Closely with Clinical Practice
	Subtheme 2: The Need for Nursing Interventions to Be More Aligned with Clinical Practice and Patient-Specific Characteristics
	Subtheme 3: The Need for Improvement in the Scheduling of Nursing Interventions to Align with Clinical Applicability
	Subtheme 4: Nursing Intervention Scheduling Should Utilize Multiple Intelligent Methods to Capture Relevant Patient Information for Documentation
Theme 2: Experiences and Optimization Needs Regarding “Right People”	None
Theme 3: Experiences and Optimization Needs Regarding “Right Formats”	Subtheme 1: Nursing Records Need to Be Integrated to Present the Patient’s Condition More Directly and Clearly
	Subtheme 2: Shift Handover Summary Records Need to Be Simplified for Easier Review
	Subtheme 3: The Format of Notifications Needs to Integrate with the Primary Clinical Workflow
Theme 4: Experiences and Optimization Needs Regarding “Right Channels”	Subtheme 1: Mobile Devices Need to Be More Usable and Compatible
	Subtheme 2: Platform Synchronization Needs to Be Seamless Between Mobile and Desktop Platforms
Theme 5: Experiences and Optimization Needs Regarding “Right Time”	Subtheme 1: System Latency Needs to Be More Effectively Minimized for Timely Decision-Making
	Subtheme 2: Clinical Reminders Need to Be More Precise Throughout Their Lifecycle
Theme 6: Experiences and Optimization Needs Regarding “Network and Technical Support”	Subtheme 1: Improving the Stability of the System Network
	Subtheme 2: Enhancing the Inference Engine’s Ability to Utilize Real-Time Data
	Subtheme 3: Improving the Responsiveness of Technical Support
Theme 7: Enhancing System Effectiveness by Standardizing Nursing Protocols for Regulating Nurse Behavior	None

3.3. Theme 1: Experiences and optimization needs regarding “right information”

3.3.1. Subtheme 1: The need for nursing diagnoses to align more closely with clinical practice

The system automatically suggests nursing diagnoses based on patient assessments using a reasoning engine. However, the recommendations often do not align with clinical practice or reflect the patient’s actual condition. Participants expressed uncertainty about the basis for these suggestions, which were seen as inconsistent with clinical reality. Consequently, nurses frequently made manual adjustments, highlighting a critical need for algorithmic refinement.

For example, “lack of knowledge” was commonly identified by nurses as a primary diagnosis, particularly upon patient admission. However, the system did not proactively suggest this diagnosis, and nurses typically needed to input it manually.

Participants emphasized the need for a more intelligent system capable of synthesizing a patient’s medical history to recommend highly relevant nursing issues.

3.3.2. Subtheme 2: The need for nursing interventions to be more aligned with clinical practice and patient-specific characteristics

The knowledge base underlying the system is based on textbooks and clinical guidelines, lending it significant scientific validity. However, since it uses standardized terminology (Clinical Care Classification, CCC) derived from descriptive records from home care institutions, it does not always align with clinical practices in domestic tertiary hospitals. The system’s recommendations sometimes conflict with hospital protocols, limiting their utility for daily decision support. Key issues include a fixed intervention model that overlooks individualized patient care and a lack of specialized recommendations, both of which hinder the delivery of targeted nursing interventions.

3.3.3. Subtheme 3: The need for improvement in the scheduling of nursing interventions to align with clinical applicability

Nursing intervention scheduling involves the planning and assignment of specific tasks based on predetermined frequencies. While the system sets default frequencies according to clinical consensus, it does not tailor these recommendations to each patient’s individual condition. Nurses are required to manually adjust intervention frequencies, which increases their operational workload.

Moreover, the system assigns tasks uniformly without distinguishing between day and night shifts. This lack of differentiation exacerbates workload imbalances, particularly during night shifts when nurse-to-patient ratios are lower and certain routine interventions are typically not performed.

Respondents noted that nurses prioritize documenting critical nursing actions, whereas tasks like health education are often omitted if already performed. However, the system mandates recording all interventions, reducing time available for direct patient care.

3.3.4. Subtheme 4: Nursing intervention scheduling should utilize multiple intelligent methods to capture relevant patient information for documentation

As a clinical decision support system, the system should integrate key patient data—such as medical orders, test results, and examinations—to trigger appropriate nursing interventions and align the nursing process with clinical workflows. For example, it should recognize surgical procedures and automatically pause vital sign monitoring entries during such events.

Due to limitations in data capture for intelligent notifications, fully automated interventions remain challenging. While the Nu-CDSS provides essential alerts, nurses must still rely on clinical expertise and manual judgment for final decisions. The main burden, however, lies in the repetitive manual data entry and administrative selections required to bypass irrelevant prompts, which detract from true clinical reasoning and reduce user-friendliness.

3.4. Theme 2: Experiences and optimization needs regarding “right people”

The system interfaces with the hospital’s scheduling platform, allowing head nurses to assign beds responsibilities to the designated nurses on a daily basis. Nurses can access and operate only on patients under their charge, ensuring relevant information reaches the appropriate person. This design enhances data security and task clarity, and most respondents confirmed it functions effectively.

However, some nurses suggested prioritizing reminder notifications. High-priority alerts could trigger full-screen prompts to ensure visibility, promoting timely responses and facilitating team coordination through mutual reminders.

3.5. Theme 3: Experiences and optimization needs regarding “right formats”

3.5.1. Subtheme 1: Nursing records need to be integrated to present the patient’s condition more directly and clearly

Unlike traditional paper records that integrate all observations chronologically, the system separates documentation into forms by intervention type. While this enhances clarity, it creates inefficiencies when nurses must record multiple observations—such as wounds and catheters—after surgery, requiring frequent tab switching. This segmentation increases operational burden and slows documentation in time-sensitive scenarios.

3.5.2. Subtheme 2: Shift handover summary records need to be simplified for easier review

The system can automatically generate shift handover summary records by extracting and aggregating patient data. While, for clinical nurses, these automatically extracted records are often lengthy and contain a large amount of irrelevant information. Consequently, nurses must spend additional time searching for key details within extensive texts, which may lead to information overload and cognitive fatigue.

3.5.3. Subtheme 3: The format of notifications needs to integrate with the primary clinical workflow

The current format of notifications, relying on a speaker-shaped icon on the login homepage and flashing lights on electronic patient panels, disconnects from the nurses’ primary active workspace. In clinical practice, responsible nurses predominantly focus their attention on the intervention scheduling interface. Because this current notification format forces nurses to navigate away from their primary workflow, usage rates remain low, and critical reminders can be easily missed.

To resolve this, future iterations of the Nu-CDSS must ensure that alert notifications are embedded directly within the intervention scheduling interface, allowing nurses to receive and respond to reminders without disrupting their established clinical workflow.

3.6. Theme 4: Experiences and optimization needs regarding “right channels”

3.6.1. Subtheme 1: Mobile devices need to be more usable and compatible

Commonly used mobile devices in clinical practice include Personal Digital Assistants (PDAs) and other handheld devices. However, mobile compatibility requires further enhancement to improve bedside usability. Participants noted that PDAs have a remarkably low usage frequency because the system’s extensive content does not translate well to a small, lag-prone interface. Consequently, nurses rarely use them, leading to a waste of hardware resources. While handheld devices are primarily used for bedside interventions, they are often unable to seamlessly load the full scope of the system, limiting the effectiveness of this channel.

3.6.2. Subtheme 2: Platform synchronization needs to be seamless between mobile and desktop platforms

A critical barrier to clinical efficiency is the fragmented data flow between bedside mobile devices and the desktop Nurse Information System. Although handheld devices are essential for bedside identity verification and executing interventions, poor platform interoperability prevents seamless data integration. Consequently, completed bedside interventions lack stable platform synchronization with the Nu-CDSS, and errors executed on mobile devices often cannot be undone locally due to cross-platform data silos. Furthermore, the lack of comprehensive system integration forces nurses to frequently switch between the Nu-CDSS and legacy systems for specific tasks, such as surgical handovers and pain management. Participants expressed a strong need for a unified, interoperable interface that ensures reliable platform synchronization.

3.7. Theme 5: Experiences and optimization needs regarding “right time”

3.7.1. Subtheme 1: System latency needs to be more effectively minimized for timely decision-making

Timely acquisition and processing of information are critical to ensuring that nurses have access to the latest patient data. To achieve this, the system’s internal processing must minimize data latency, ensuring information is presented to decision-makers exactly when needed. Excessive system latency and delayed data processing fail to reflect newly acquired clinical values promptly, which compromises the accuracy and effectiveness of immediate clinical decisions.

3.7.2. Subtheme 2: Clinical reminders need to be more precise throughout their lifecycle

The entire lifecycle of clinical reminders, from initiation to resolution, must perfectly align with the patient’s actual condition. Participants highlighted two critical temporal phases requiring optimization. First, trigger precision must be enhanced: the system needs to push reminders based on physiological changes with precise timing, rather than relying on fixed time ranges. Second, timing optimization must apply to reminder removal. Once a clinical task is executed, the system must promptly remove the associated reminder marker. Delayed removal leads to prolonged screen clutter, information overload, and alert fatigue.

In addition to the factors covered by the 5R theory, the interviewees also identified technical support, network stability, and the establishment of behavioral standards as other crucial factors affecting the system’s user experience, which also require optimization.

3.8. Theme 6: Experiences and optimization needs regarding “network and technical support”

3.8.1. Subtheme 1: Improving the stability of the system network

The speed and stability of the system’s network are critical factors affecting the user experience. An unstable network slows down the system’s operation, increases wait times, and directly impacts the user experience. This, in turn, indirectly reduces the time nurses can spend providing direct bedside care. In the current interviews, slow network speed was consistently identified as a major issue.

3.8.2. Subtheme 2: Enhancing the inference engine’s ability to utilize real-time data

The inference engine, or the system’s decision-making rules, is a crucial component of a clinical decision support system. If the inference engine cannot effectively utilize real-time data or adjust to new information and changing circumstances, it limits the system’s ability to respond to emergent clinical situations. This can lead to the system providing nursing schedules that do not align with current clinical realities, thus increasing the operational burden on clinical staff.

3.8.3. Subtheme 3: Improving the responsiveness of technical support

In daily practice, ward nursing staff can contact technical support through various channels. However, due to the limited capacity of technical personnel, nurses’ improvement requests are often not addressed promptly. This delays the system’s optimization and decreases nurses’ enthusiasm to provide feedback.

3.9. Theme 7: Enhancing system effectiveness by standardizing nursing protocols for regulating nurse behavior

The Nu-CDSS system’s built-in knowledge base includes numerous standardized nursing measures, some of which lack clear directives, causing confusion during intervention selection and frequency determination during care plan development. To resolve these uncertainties, nurses typically rely on informal peer guidance or department-specific protocols established by managers. Consequently, this reliance on localized workarounds leads to inconsistent nursing practices and fragmented system utilization across different departments.

4. Discussion

This study explored the experiences and optimization needs of clinical nurses and nurse managers who directly interact with the Nu-CDSS. Seven key themes were identified, highlighting priorities for improving the knowledge base, human-computer interface, inference engine, and network infrastructure. Addressing these areas is essential to enhance system effectiveness, streamline clinical workflows, and improve patient outcomes.

4.1. The need for a refined knowledge base to deliver the right information at the right time

The knowledge base is a core component of the Nu-CDSS. ¹⁸ While routine risk recommendations, such as falls and pressure injuries, are generally accurate, broader clinical adaptability and personalization are still required. However, discrepancies exist between the system-recommended nursing diagnoses and actual clinical practice. ⁶ In China, domestic nursing diagnoses are primarily derived from textbooks aligned with North American Nursing Diagnosis Association (NANDA) standards. ¹⁹ Because diagnoses are deeply rooted in clinical experience, static system recommendations frequently conflict with nurses’ professional judgment. To improve accuracy and clinical relevance, the Nu-CDSS must continuously integrate user feedback and expert consensus to refine its diagnostic recommendations. ²⁰

Participants noted that system-recommended interventions occasionally diverge from local hospital protocols, limiting nurses’ reliance on the Nu-CDSS for daily decision-making. ²¹ While default frequencies align with guidelines, manual adjustments to maintain nursing autonomy create discrepancies between planned and actual care. Integrating nurses’ patient-specific selections is essential to iteratively refine the knowledge base.

This rigidity reflects global challenges in traditional rule-based Nu-CDSS deployments, necessitating a shift toward AI-enhanced systems. The Nu-CDSS should incorporate real-time multidimensional data, applying machine learning and Natural Language Processing (NLP) to dynamically localize its knowledge base. ²² Ultimately, AI-driven analytics can bridge standardized terminologies with daily nursing realities, ¹ optimizing complex decision-making support.

4.2. Enhancing the human-computer interface to optimize the right formats and right channels

The formatting and organization of Nu-CDSS information are critical to its clinical usability. To optimize the human-computer interface, the system must integrate intuitive visual icons and task-oriented layouts to simplify navigation and improve data access. Tailoring these “Right Formats” to routine nursing workflows simplifies navigation, accelerates decision-making, and minimizes errors. ²³

Given the dynamic and frequently interrupted nature of nursing work, optimizing “Right Channels” requires flexible pause, save, and resume capabilities. Incorporating undo/redo functions and confirmation prompts for irreversible actions is also essential to enhance error management, preserve task progress, and maintain system reliability in fast-paced clinical environments. ²⁴

Nursing records and handover summaries are heavily text-based, requiring nurses to process extensive information and increasing cognitive workload. This finding echoes previous research highlighting the importance of streamlining information presentation to prevent cognitive overload. ²⁵

4.3. Optimizing the inference engine to enhance timely decision-making

The inference engine is vital for Nu-CDSS decision-making, yet participants indicated that it adapted poorly to real-time data and evolving clinical conditions. Its reliance on fixed algorithms and a static knowledge base limited its ability to reflect dynamic patient needs, resulting in diagnosis suggestions, intervention schedules, and reminders that did not always align with clinical reality. ²⁶ These findings suggest a need for more adaptive and context-sensitive decision logic.

In this study, AI was not viewed as a generic future upgrade, but as a targeted response to specific deficits identified by nurses. These included improving the localization of diagnosis recommendations, tailoring intervention scheduling to patient condition and shift context, and strengthening the precision of reminder triggering and removal. By embedding machine learning, predictive modeling, and natural language processing, the system may better capture multidimensional clinical data and provide more accurate, context-sensitive support. ²⁷ This direction also aligns with growing recognition among nurses that AI can enhance practice when integrated into routine workflows. ²⁸

Importantly, AI-enhanced Nu-CDSS should support rather than replace nurses’ clinical judgment. The goal is to reduce redundant administrative work and improve the relevance of alerts and recommendations, while preserving nurses’ capacity for complex reasoning, patient assessment, and individualized care. ² In this way, future Nu-CDSS may function more effectively as a symbiotic partner in clinical decision-making.

4.4. The need for timely and continuous technical support and network optimization

Professional technical support is essential for the stability, optimization, and adaptability of the Nu-CDSS. ²⁹ A skilled support team enhances system reliability by efficiently diagnosing issues, resolving errors, and refining functionality based on user feedback, ³⁰ thereby supporting optimal performance and confident clinical decision-making. Because nurses heavily depend on the system, technical malfunctions directly disrupt clinical efficiency, driving a strong demand for agile support. ³¹ Timely resolution of clinical feedback is vital for system optimization. Establishing a dedicated IT team and an active user feedback mechanism can ensure timely resolution of clinical issues and facilitate continuous system evaluation and improvement.

Network stability is also a key focus in optimizing the Nu-CDSS. It is vital for data transmission, remote access, information security, and system reliability. ³² Network instability compromises data timeliness, disrupts nursing workflows, and increases workload. Addressing these challenges requires leadership-backed investments in infrastructure, bandwidth, and security to restore decision-support efficiency. As technical capacities expand, the ethical implications of processing sensitive clinical data, such as continuous vital sign and behavioral tracking, must also be addressed. Future Nu-CDSS optimization must go beyond functional utility to prioritize patient privacy through robust data encryption, strict access controls, and transparent data-use protocols.

4.5. Implications for clinical nursing practice

Optimizing a Nu-CDSS extends beyond technical usability, requiring fundamental adaptations within the clinical environment. Superimposing digital systems onto traditional routines causes redundant documentation and workflow disruptions. Therefore, hospital administrators must proactively restructure workflows, ensuring clinical processes and system logic are mutually aligned and co-optimized through standardized procedures. Additionally, incorporating system enforceability at critical points is vital to mitigate human errors and safeguard patient safety.

Furthermore, beyond basic IT onboarding, continuous, scenario-based training is essential to elevate nurses’ digital literacy and critical thinking. This empowers nurses to transition from passive reliance on system prompts to engaging the CDSS as an active partner in clinical reasoning, ultimately translating technological adoption into tangible improvements in patient care.

4.6. Limitations

This study has several limitations. As a small qualitative study conducted in a single tertiary hospital in Shanghai, involving a select, stratified group of Chinese-speaking nurses, its findings may have limited transferability to other clinical settings. Moreover, all participants were female, reflecting the local nursing workforce. Although the findings are not broadly generalizable, they offer useful context-specific insights for optimizing Nu-CDSS in acute care practice.

5. Conclusion

Guided by the 5R framework, this study identified seven themes reflecting nurses’ priorities for optimizing the Nu-CDSS. Key needs included a more localized and adaptive knowledge base, better workflow-integrated interfaces, improved platform synchronization and reminder timing, and stronger technical and network support. Future iterations should move toward AI-enhanced Nu-CDSS that provide more adaptive and context-sensitive support while preserving nurses’ central role in clinical judgment. These findings provide context-specific guidance for refining Nu-CDSS in acute care settings.

Supplemental material

Supplemental material - Optimizing nursing clinical decision support systems: Insights from nurses’ experiences

Supplemental material for Optimizing nursing clinical decision support systems: Insights from nurses’ experiences by Qi Zhang, Aliaya·Adili, Zhenghong Yu, Wei Qin, Zheng Zhu, Yuxia Zhang in Health Informatics Journal