Healthcare professionals’ perspectives on artificial intelligence (AI)-based mobile applications (apps) for diabetes education and behavioural management

Abstract

Background

The global prevalence of diabetes mellitus is escalating rapidly. Similarly in Malaysia, diabetes prevalence among adults rose exponentially. Artificial intelligence (AI) integrated into mobile health applications (apps) presents a promising avenue for enhancing diabetes management through personalized patient education and behaviour modification. However, adoption rates remain low, primarily due to limited awareness and technological challenges, especially among older adults.

Objective

This study aimed to explore Malaysian healthcare professionals’ (HCPs’) perspectives on the use of AI in mobile apps for diabetes education and behavioural management.

Methods

Qualitative study using semi-structured interviews with 19 HCPs across Malaysia. Interviews were conducted via video conferencing, recorded, transcribed, and analysed using thematic analysis with NVivo 14 software.

Results

Seven key themes emerged: (1) acceptance and trust, (2) impact on patient behaviour, (3) skills and abilities required, (4) problems and obstacles, (5) key features and functions, (6) HCPs’ and patients’ information needs, and (7) strategies for increasing patient adoption. HCPs expressed positive sentiments towards AI-based apps, highlighting their potential for continuous, personalized education and real-time feedback. However, significant concerns were raised about accessibility for older adults, data privacy, and the apps’ ability to modify entrenched behaviours without human intervention.

Conclusion

AI-based mobile apps show potential for improving diabetes management, but successful implementation requires addressing challenges. Strategies should focus on developing user-friendly interfaces, providing comprehensive education for patients and providers, and ensuring robust data protection. Future research should quantify the impact on patient outcomes and explore effective integration of human support with AI capabilities.

Keywords

Artificial intelligence diabetes mellitus mobile applications patient education behaviour change

Introduction

Diabetes mellitus (DM) represents a significant global health challenge, with the International Diabetes Federation reporting that approximately 537 million adults are currently affected, and projections suggest this number will increase to 643 million by 2030.¹ The impact of this epidemic is particularly pronounced in Malaysia, where the prevalence of diabetes among adults increased by 68% from 2011 to 2019.² This rapid increase not only affects individual health outcomes, but also places a substantial burden on healthcare systems and economies worldwide.

In recent years, the integration of artificial intelligence (AI) into mobile health applications (apps) has emerged as a promising approach to improve diabetes management, particularly through personalized patient education and behaviour modification.³ These AI-based apps offer the potential to provide continuous support, real-time feedback, and tailored interventions that extend care beyond traditional clinical settings.^4–10 However, despite their potential benefits, the adoption of these technologies remains low, especially among older adults who often face challenges related to digital literacy and access.¹¹

Previous research has explored various aspects of mobile health interventions for diabetes management, including their effectiveness in improving glycaemic control,¹² improving medication adherence,^13–16 and promoting lifestyle modifications. However, there is a paucity of research examining healthcare professionals’ (HCP’) perspectives on the integration of AI into these mobile apps, particularly in the context of developing countries like Malaysia.¹⁷

Understanding the views, experiences, and concerns of HCPs is crucial for the successful implementation and adoption of AI-based mobile apps in diabetes care. These insights can inform the development of more effective and user-friendly apps, guide the creation of targeted training programs for both healthcare providers and patients, and help address potential barriers to adoption.^18,19

Therefore, this study aimed to explore the perspectives of HCPs in Malaysia regarding the use of AI in mobile apps for diabetes education and behavioural management. By elucidating these viewpoints, we seek to contribute to the growing body of knowledge on digital health interventions^20,21 and provide valuable insights for policymakers, healthcare organizations, and technology developers working to improve diabetes care through innovative digital solutions.²²

Methodology

Study design and participants

This qualitative study employed semi-structured interviews to explore HCPs perspectives on the use of AI in mobile apps for diabetes education and behavioural management. The study adhered to the International Conference on Harmonization and Good Clinical Practice guidelines and was approved by the Medical Research Ethics Committee, Ministry of Health, Malaysia (No. NMRR-23-01186-APO) and the Universiti Tunku Abdul Rahman’s Scientific and Ethical Review Committee (U/SERC/27/2023).

The participants were HCPs from different healthcare organizations in Malaysia. Participants were recruited using purposive sampling to ensure a diverse range of HCPs across various specialities and geographic locations in Malaysia. The researchers approached potential participants personally, offering information about the study. Study invitations were distributed to potential participants. All approached interviewees agreed to participate in the study. This recruitment strategy allowed for a comprehensive representation of healthcare perspectives from different regions and specialities within Malaysia. The inclusion criteria were:

HCPs of any gender,

aged between 21 and 70 years,

experience in managing DM patients, and

current employment in either public or private healthcare facilities.

HCPs with cognitive impairment or psychiatric conditions were excluded.

A total of 19 HCPs participated in the study, representing a diverse range of specialities and geographic locations across Malaysia. The sample included eight pharmacists, four endocrinologists, three family medicine specialists, two dieticians, and two diabetes educators. Participants were recruited from public health facilities $(n = 12)$ , university hospitals $(n = 4)$ , and private health facilities $(n = 3)$ , representing 10 states in Malaysia.

Data saturation was carefully monitored throughout the data collection process. After conducting interviews with 15 participants, we observed that no new themes or significant insights were emerging from the data. To ensure saturation had indeed been reached, we conducted four additional interviews, which confirmed that data saturation had been achieved. This approach aligns with established practices in qualitative research for determining sample size adequacy.^23–26

Their demographic characteristics are presented in Table 1.

Table 1.

Demographic data of the participants.

Participant	Age (years)	Ethnicity	Specialty	Working experience (years)	Experience with diabetes patients (years)
R01	37	Chinese	Pharmacist	13	10
R02	42	Chinese	Diabetes educator	15	12
R03	36	Chinese	Pharmacist	13	12
R04	41	Chinese	Pharmacist	16	14
R05	43	Chinese	Family medicine Specialist	17	17
R06	45	Chinese	Dietitian	20	20
R07	43	Indian	Family medicine specialist	18	13
R08	37	Malay	Pharmacist	13	7
R09	48	Malay	Endocrinologist	>20	>20
R10	46	Chinese	Endocrinologist	21	21
R11	39	Chinese	Endocrinologist	15	13
R12	50	Indian	Endocrinologist	23	10
R13	38	Chinese	Pharmacist	15	14
R14	35	Chinese	Dietician	11	11
R15	33	Malay	Pharmacist	14	14
R16	30	Malay	Pharmacist	7	6
R17	39	Chinese	Pharmacist	16	7
R18	39	Chinese	Family medicine specialist	14	14
R19	69	Chinese	Diabetes educator	47	36

Data collection

Data were collected using semi-structured individual interviews. All questionnaires used in this study were originally developed by the authors for the specific purposes of this research. No external copyrighted materials requiring permission for use were utilized in this study. The interview protocol was developed through a rigorous process:

Comprehensive literature review

An initial review of databases including IEEE, ACM, Springer, Wiley, Sage and Elsevier yielded 78 articles on diabetes, AI, and mobile applications. Of these, 29 studies included questionnaires about HCPs’ perspectives on mobile applications in diabetes management. After screening, seven relevant studies were identified, and 53 potential questions were initially considered.^27–33

Nominal group technique

Five experts, including computer science lecturers, a pharmacist specializing in diabetes care, and a psychologist, refined the questions. After five rounds of discussions, 45 redundant questions were excluded, leaving eight distinct questions.

Content validity

Additional experts, including an HCP from the Ministry of Health and an academic researcher specializing in diabetes management, validated the question. Each question was rated on a 4-point scale for relevance, importance, and clarity to calculate the content validity index (CVI). The item-level content validity index (I-CVI)³⁴ was calculated to establish the validity of the interview questions using the equation (1):

\frac{I-CVI = Number of experts who rate 3 or 4}{Total number of experts}

(1)

Content validity is considered excellent if the I-CVI value is at least 0.80, with evaluations carried out by two experts.³⁵

Final refinement

Based on expert feedback, the questions were further refined. The second draft of the questions underwent content validation again and I-CVI was calculated. The final set comprised eight open-ended questions in English, considering the proficiency of Malaysian HCPs in the language.

The final interview protocol included:

A 5-minute ice-breaking session to build rapport.

Eight open-ended questions addressing AI in diabetes education, behavioural management, application barriers, app features, and data needs.

The final set of interview questions are as follows:

What is your opinion about the idea of using AI in mobile apps to educate patients about diabetes management?

What is your opinion about the idea of using AI in mobile apps to change patient behaviour in diabetes management?

Which skills and abilities should the patients have when using AI-based mobile apps to educate themselves and change their behaviour in diabetes management?

What potential problems or obstacles should be considered when patients use AI in mobile apps to educate themselves and change their behaviour in diabetes management?

How should patients access AI-based mobile apps related to diabetes?

Could you identify the features and functions that AI-based diabetes management mobile apps should have to increase their adoption by patients?

What data or information do you think the patients need from AI-based mobile apps to educate themselves and change their behaviour?

What data or information do you wish to collect from patients when using AI-based mobile apps to educate themselves and change their behaviour?

The interviews were conducted by three researchers (first author (PC), second author (YW), and fifth author (KL)) between May and June 2023 via Webex video conferencing. PC is a pharmacist specializing in diabetes care, whereas YW and KL are computer science lecturers at their respective universities. There was no prior relationship between the interviewers and the interviewees. Each interview lasted approximately 30 minutes. Participants were not previously informed about the content to minimize bias. All sessions were recorded with participant consent, and verbatim transcriptions were made. Notes were taken during interviews, and all data were securely stored, accessible only to the research team.

Data analysis

All interviews were transcribed verbatim and analysed using thematic analysis facilitated by NVivo 14 software. The analysis process followed the six-phase thematic analysis approach,³⁶ detailed in Figure 1.

Familiarization with the data: This involved reading and re-reading the transcripts, noting down initial ideas.

Generating initial codes: Systematic coding of interesting features across the entire data set, collating data relevant to each code.

Combining codes into themes: Gathering all data relevant to each potential theme.

Reviewing themes: Checking if the themes work in relation to the coded extracts and the entire data set, generating a thematic ‘map’ of the analysis.

Determine significance of themes: Ongoing analysis to refine the specifics of each theme, and the overall story the analysis tells, generating clear definitions and names for each theme.

Producing the report: Final opportunity for analysis. Selection of vivid, compelling extract examples, final analysis of selected extracts, relating back of the analysis to the research question and literature, producing a scholarly report of the analysis.

Figure 1.

Six-phase thematic analysis.

This rigorous process ensured a comprehensive and systematic analysis of the data. The coding structure and emerging themes were regularly reviewed by the entire research team to ensure rigour and trustworthiness. The final themes and subthemes were summarized in several hierarchy charts and mind maps to visually represent the findings.

To quantify the prevalence of these viewpoints, percentages were calculated based on the frequency of each perspective expressed across all participants. It is important to note that these percentages do not represent mutually exclusive categories, as individual participants often expressed multiple viewpoints within a single theme. This approach allows for the representation of the relative prominence of different perspectives while acknowledging the complex and multifaceted nature of HCPs’ views on AI-based mobile apps for diabetes management.

Results

Item-level content validity index (I-CVI)

In the first evaluation by both experts, the eight questions were graded as relevant and important with an I-CVI of 1.0. Questions numbers 2, 4, and 6 had low clarity with an I-CVI of 0.5. Modifications to the questions based on experts’ comments included splitting questions for education and behavioural change. The second draft of the interview questions also consisted of eight open-ended questions. The second content validation showed that the questions were relevant, important, and clear with an I-CVI of 1.0, except for question number 5, which had an I-CVI of 0.5 for clarity. This question was then modified.

Data analysis

The thematic analysis of 19 semi-structured interviews with HCPs in Malaysia revealed seven key themes on the use of AI-based mobile apps for diabetes education and management. These themes are summarized in Figure 2 and include: (1) acceptance and trust, (2) impact on patient behaviour, (3) skills and abilities, (4) problems and obstacles, (5) key features and functions, (6) HCPs’ and patients’ information needs, and (7) strategies for enhancing patient’s engagement with AI-mobile apps.

Figure 2.

Summary of themes.

It is crucial to interpret these figures with the understanding that the percentages represent the prevalence of each viewpoint across all participants, rather than mutually exclusive categories. Individual HCPs often expressed multiple, sometimes overlapping viewpoints within each theme, reflecting the nuanced and complex nature of their perspectives on this technology.

Acceptance and trust

HCPs generally expressed positive sentiments towards AI-based mobile apps for diabetes education. The main factors contributing to acceptance included increased accessibility to information and care, effectiveness in education, and improved patient learning. As illustrated in Figure 3, 35.29% of the participants found mobile AI apps effective for learning about diabetes management, while 23.53% emphasized their effectiveness in increasing accessibility and patient learning.

Figure 3.

Distribution of HCPs’ perspectives on acceptance and trust in using AI-based mobile apps. HCP: healthcare professional; AI: artificial intelligence.

Accessibility emerged as a key factor, allowing patients to access diabetes management information without relying solely on physical visits to their healthcare workers (HCWs). R02 elaborated on this point:

It is a very good powerful supporting tool in diabetes education, especially for patients who can’t access a diabetes educator or DMTAC pharmacist. For hospitals without these services, dedicated patients can find support through these mobile apps. It is very convenient.

This sentiment was reinforced by R09, who noted:

It’s the way forward because it can reach many patients simultaneously. They can use it at their convenience, with different modules. It’s a very good approach.

The effectiveness of these apps in educating patients about diabetes management was another crucial aspect. R12 observed:

It will make patients take ownership of their own self-management, likely making it more effective.

R05 further highlighted the potential of AI in supporting diagnosis:

For junior people, AI can help make better diagnoses. For diabetes, AI can remind you to do necessary checks like eye and foot examinations.

The apps’ effectiveness in increasing information access through additional lessons improves care, as noted by R13:

We don’t see patients daily or even weekly. At most, it’s monthly or every three months. That’s not enough to educate them sufficiently. A mobile app would give them more time to learn daily. That would be more effective.

In general, HCPs perceived AI-based apps for patient education to be a valuable and powerful tool, particularly to improve accessibility to diabetes education. R18 summarized this sentiment:

Education is crucial in diabetes management. At the clinic, we have limited time for education. Patients often don’t retain information from consultations, so it’s good to have something outside the clinics.

Impact on patients’ behaviour

This theme examines how AI-mobile education apps can influence and potentially change patient behaviour towards diabetes management. Views on the potential of AI-based mobile apps to influence patient behaviour were mixed. As shown in Figure 4, 36.36% of the participants believed that AI-mobile apps motivate healthier behaviour habits in patients, 27.27% found these apps effective in increasing self-care and monitoring, while another 27.27% highlighted the difficulty of behaviour change.

Figure 4.

Distribution of HCPs’ perspectives on impact on patients’ behaviour in using AI-based mobile apps. HCP: healthcare professional; AI: artificial intelligence.

Many HCWs expressed hope that AI-mobile apps would motivate patients’ healthier behaviour habits by serving as a constant source of information for diabetes management. R07 explained:

AI will be a motivational tool to change behaviour. Constant motivation is key for behavioural change, and AI can provide that continuous motivational input.

R15 highlighted the potential of AI-mobile apps to motivate patients through monitoring and data tracking:

The mobile app can assist patients in changing behaviour because we can monitor them. Patients have all the information, so they become aware and change accordingly. AI can track patient data, efficiently changing their behaviour.

Some participants asserted that AI-mobile apps made it easier for patients to change their behaviour by providing an opportunity to capture a snapshot of their habits and health. R14 elaborated:

The system should share trends in behaviours, like tracking mental health status and correlating it with eating habits and blood glucose. This helps patients think about lifestyle changes and come up with solutions to reinforce behavioural change.

The increased self-care and monitoring facilitated by AI-mobile apps was seen as an essential element in positively impacting patient behaviour. R04 explained:

AI in mobile apps are powerful for people with diabetes as they can take control of their own management. AI can help them make decisions about activity and diet by providing lots of information.

However, a significant portion of participants (27.27%) acknowledged the difficulty associated with attaining patient behaviour change despite using AI-mobile apps, primarily due to the lack of human interaction. R03 highlighted this concern:

I’m not sure if it will really help change patient behaviour. I feel personal touch is still important.

Skills and abilities

HCPs identified several key skills patients need to effectively use AI-mobile apps. As illustrated in Figure 5, 40% of participants emphasized literary and research skills, 30% highlighted social and language skills, 20% mentioned technical know-how, and 10% noted self-efficacy and motivation as key factors for effective use of AI-mobile apps.

Figure 5.

Distribution of HCPs’ perspectives on key skills patients need to effectively use AI-based mobile apps. HCP: healthcare professional; AI: artificial intelligence; apps: applications.

Literary and research skills were identified as the fundamental requirement. This includes the ability to read and understand instructions, interpret graphs, and find validated resources. R03 explained:

Patients need literacy in technology. They should be able to read, understand instructions, and interpret graphs generated by the app.

R08 emphasized the importance of research skills:

They need to find validated resources, not just follow any advice. They should know how to find expertise for validated advice, not just marketing.

Social and language skills were the second most important, highlighted by 30% of participants. This includes the ability to communicate effectively and use the app in their preferred language. R02 highlighted the need for effective communication:

Patients need to communicate or identify their problems to input into the app. This allows the app to personalize treatment.

R10 emphasized the importance of language skills:

They need language skills to read, listen, and understand the information given.

Technical know-how, which includes the ability to download apps, enter data, and operate smartphones, is also important. R14 summarized the required technical skills:

Basic technology skills are needed to navigate the apps, use them, and insert data.

R16 further emphasized this point:

They need good technical skills for handling mobiles and gadgets. If they can manoeuvre through the app, they’ll be okay using it.

Finally, self-efficacy and motivation were noted by 10% of participants as crucial for the effective use of AI-mobile apps. R02 explained:

The most difficult thing is that patients need self-motivation or self-discipline to follow the mobile app’s instructions.

Problems and obstacles

This study identified several problems and obstacles associated with using AI-mobile apps for patient education in diabetes management. As illustrated in Figure 6, the main challenges can be categorized into five areas: technology problems (30%), need for human participation (30%), data privacy and protection (20%), language barriers (10%), and information reliability/accuracy (10%).

Figure 6.

Distribution of HCPs’ perspectives on problems and obstacles in implementation of AI-based mobile apps. HCP: healthcare professional; AI: artificial intelligence.

The technology problems identified included unfriendly user interfaces, poor access, and inadequate internet connectivity. R16 emphasized the need for a straightforward app design:

App should be straightforward, not too many buttons or navigations for updates.

R04 highlighted accessibility issues, particularly in rural areas:

For urban patients, access should be no problem. But for rural areas with more districts, it may not be applicable for some patients.

Participants also emphasized the need for human participation, expressing concern about the lack of direct human interaction that is common between patients and HCWs. R03 articulated this concern:

I support AI mobile apps, but I’m skeptical if they can achieve the human touch. We build connections with patients by asking about their lives. This rapport helps change behaviours in diabetes management, which AI might struggle with.

Given the sensitive nature of patient medical records, R14 raised the issue of data privacy and protection:

Privacy is a concern. If patients feel unsafe using the apps or recording data due to security issues, we need to reinforce privacy measures.

R13 explained that language differences can also become a barrier:

We must consider language barriers. Someone who doesn’t understand much English might have problems if the app doesn’t include their mother tongue.

Concerns about the reliability of the information were expressed by R10:

I worry about inaccurate information. This can happen if inaccurate information is fed to the AI, like patients not giving truthful information.

Key features and functions

HCPs suggested several essential features and functions for AI-based mobile apps to enhance patient adoption in diabetes management. As illustrated in Figure 7, the most prominent features include an easy and friendly user interface (33.33%), technical features (16.67%), clear instructions (16.67%), and patient self-monitoring functions (16.67%). Privacy and security, as well as reminders and alerts, were each highlighted by 8.33% of participants.

Figure 7.

Distribution of HCPs’ perspectives on essential features for AI-based mobile apps. HCP: healthcare professional; AI: artificial intelligence.

R13 emphasized the importance of user-friendliness:

It should be very easy to use – user friendly.

R16 elaborated on the importance of simple navigation:

Navigating should be easy, with just one click to update.

R10 suggested incorporating interactive elements to enhance engagement:

Interactive elements like simple games can attract attention. Some apps just feed information, but a TikTok-style video could catch patients’ interest.

Technical features include AI flexibility, Bluetooth connectivity, and integration with wearable technology. R07 emphasized the need for adaptive AI:

The AI should adapt itself to the patient’s needs.

R14 explained the importance of the integration of the wearable device:

Integration with wearable devices is important. For tracking patient activity, wearable devices are very helpful.

R02 emphasized the need for clear instructions and questions, as well as multi-language capability:

Instructions must be clear and understandable by lay persons, regardless of education level. It must be usable by everybody.

Patient self-monitoring functions, including real-time feedback, reminders, and alerts for diet and medication monitoring, as mentioned by R03:

I’m excited about the function where patients can use it as a carb counting tool for their diet.

R12 noted the importance of data protection:

Long-term, we need to consider legal implications of using this data. We must address data security and patient privacy.

HCPs’ and patients’ information needs

Key information needs identified for both HCPs and patients are illustrated in Figure 8. Diet information was deemed essential by 27.27% of participants, followed by blood test results (18.18%), medication and treatment management (18.18%), and lifestyle information (18.18%). Diabetes complications and overall health status were each mentioned by 9.09% of participants.

Figure 8.

Distribution of HCPs’ perspectives on key information needs for AI-based mobile apps. HCP: healthcare professional; AI: artificial intelligence.

Diet information emerged as the most crucial need, which includes carbohydrate counting and food intake monitoring. R03 explained:

Patients would benefit from carb counting tools, sugar control trend recording, and connecting glucometer readings directly to the app.

R03 indicated that HCPs and patients will need blood test results, such as glucose levels:

Also the blood test results from time to time.

R14 explained the importance of medication and treatment management:

It’s important for patients to manage their medication. What’s its function? When to take it? What if I miss a dose?

R10 added the importance of including information about medication side effects:

Patients worry about side effects. If the AI can educate them about potential side effects, that would be good.

R02 explained that lifestyle information was also deemed crucial, given the close relationship between diabetes and patients’ lifestyles:

We need information on the patient’s lifestyle habits. What are they currently practicing? Are they doing physical activity? What’s their pattern?

R12 highlighted the need for comprehensive health information, including diabetes complications and overall health status:

Their renal and liver function profile, medication adherence, and diabetes complication screening.

R05 emphasized the importance of early detection of complications:

Early detection of complications is important. Like taking a photo of the leg and sending it to us. AI can analyze for early signs of diabetic foot changes.

Strategies for enhancing patient’s engagement with AI-based mobile apps

This theme explores strategies to increase patient adoption and usage of AI-based mobile apps for diabetes management education. As illustrated in Figure 9, several key approaches were identified by HCWs.

Figure 9.

Distribution of HCPs’ perspectives on key approaches to enhance patient’s engagement with AI-based mobile apps. HCP: healthcare professional; AI: artificial intelligence.

Patient motivation emerged as the most crucial strategy, emphasized by 36.36% of participants. R17 suggested using the app to increase awareness of potential complications:

To motivate patients who feel life is meaningless if they can’t eat freely, the app can show pictures of complications like amputations.

Increasing diabetes management programs was the second most popular strategy, suggested by 27.27% of participants. R14 explained:

In Malaysia, there’s a diabetes lifestyle program with nice educational content. Patients have free consultations after completing certain modules.

Establishing rapport and fear management were each identified by 18.18% of participants as potential strategies. R16 emphasized the importance of building trust:

Building rapport with patients is crucial so they trust what we share. It’s important before we educate them, to help them understand and change behaviour.

Discussion

This study provides valuable insights into HCPs’ perspectives on AI-based mobile apps for diabetes education and behavioural management in Malaysia. The findings reveal a complex landscape of opportunities and challenges associated with the integration of these technologies into diabetes care.

The general acceptance and enthusiasm for AI-based apps among HCPs align with previous research highlighting the potential of mobile health technologies to extend care beyond traditional clinical settings. The perceived benefits of convenience, continuous access to information, and improved self-management capabilities resonate with the broader goals of digital health interventions in chronic disease management.

Our findings emphasize the importance of patient-centric education in diabetes self-care management through AI-based mobile applications. HCPs in our study highlighted several key aspects that align with empirical evidence on effective self-care management: the need for personalized education that accounts for patients’ existing knowledge levels (mentioned by 35.29%), the role of motivation in behavioural change (36.36%), and the importance of tracking preventive practices and outcomes (27.27%). These insights support previous research showing that successful intervention programs must consider patients’ knowledge base, motivational factors, attitudes towards self-care, and their ability to implement preventive practices.

However, the mixed views on the apps’ ability to effect behavioural change underscore the complex nature of health behaviour modification and the continued importance of human interaction in healthcare delivery. This tension between technological capabilities and the need for personal touch in care aligns with previous studies that have emphasized the importance of blended care models in digital health interventions.

The identification of necessary skills and abilities for effective app use highlights a potential digital divide, particularly concerning older patients or those with limited technological literacy. This finding echoes concerns raised in previous studies about the accessibility of digital health interventions across diverse patient populations. Addressing this challenge will require targeted efforts to improve digital health literacy and ensure that app designs are inclusive and user-friendly for all age groups.

The problems and obstacles identified, particularly regarding information reliability, data privacy, and the need for human interaction, reflect ongoing challenges in the field of AI in healthcare. These concerns emphasize the need for rigorous validation of AI-based health information, the development of robust data protection measures, and careful consideration of how to balance automated support with human expertise in care delivery.

The suggested features and functions for AI-mobile apps provide valuable insights for developers and healthcare organizations looking to implement these technologies. The emphasis on user-friendly interfaces, personalization, and integration with existing healthcare processes aligns with best practices in mobile health app design. Furthermore, the comprehensive information needs identified for both HCPs and patients highlight the potential for these apps to serve as integrated platforms for diabetes management, supporting various aspects of care from medication management to lifestyle modification.

Lastly, the suggested strategies for increasing patient adoption of AI-mobile apps underscore the importance of a holistic approach to implementation. This approach should incorporate patient education, motivation, and integration with existing care structures, reflecting the multifaceted nature of successful digital health interventions.

Limitations

This study has several limitations. The sample was limited to HCPs in Malaysia, which may limit the generalizability of the findings to other cultural or healthcare contexts. The study focused solely on HCP perspectives; future research should incorporate patient perspectives to provide a more comprehensive understanding of the challenges and opportunities in this field.

The interview questions underwent a rigorous development process, including expert review and content validity assessment. However, they were not previously validated or pilot-tested instruments. This could be considered a limitation in our methodology. HCPs’ perspectives may vary depending on their exposure to different AI-based mobile applications for diabetes management. Our study did not differentiate between specific AI applications or their features, which could have influenced the range of experiences and opinions expressed by participants.

Participants’ responses might have been influenced by their clinical roles, prior experiences with technology, or institutional contexts. Although we used open-ended questions to avoid leading participants, complete elimination of professional bias is not possible in qualitative research of this nature. Future research could benefit from a more targeted approach, examining experiences with specific AI tools, comparing perspectives between different types of applications, and including additional validation techniques to minimize potential biases.

While basic information on participants’ experience levels was provided, we did not assess their pre-existing attitudes towards AI-based apps (e.g. whether they were enthusiasts/early adopters or more sceptical). This limitation means that we cannot determine how these attitudes might have influenced their perspectives. Future research could benefit from a more detailed characterization of participants’ technology attitudes while still maintaining an unbiased selection process.

Lastly, as a qualitative study, our research does not provide quantitative measures of the impact of AI-based apps on patient outcomes. This is an important area that should be addressed in future research to complement our qualitative findings.

Conclusion

AI-based mobile applications show substantial potential to improve diabetes management, but their successful implementation requires addressing several key challenges. Based on the insights gained from HCPs in this study, we recommend the following strategies:

Develop user-friendly interfaces that cater to diverse patient populations, including older adults and those with limited technological literacy.

Implement comprehensive education programs for both patients and healthcare providers to enhance digital health literacy and app utilization skills.

Ensure robust data protection measures and validate the accuracy of AI-generated information to build trust among users.

Explore blended care models that effectively integrate AI support with human expertise to address the complex needs of diabetes management.

Conduct further research to quantify the impact of AI-enhanced apps on patient outcomes and explore optimal ways to integrate these technologies into existing care pathways.

By addressing these recommendations, the potential of AI-based mobile apps to enhance diabetes education and behavioural management can be more fully realized, ultimately improving patient outcomes and quality of life.

Supplemental Material

sj-pdf-1-dhj-10.1177_20552076251329991 - Supplemental material for Healthcare professionals’ perspectives on artificial intelligence (AI)-based mobile applications (apps) for diabetes education and behavioural management

Supplemental material, sj-pdf-1-dhj-10.1177_20552076251329991 for Healthcare professionals’ perspectives on artificial intelligence (AI)-based mobile applications (apps) for diabetes education and behavioural management by Phei-Ching Lim, Yung-Wey Chong, Qiu-Ting Chie, Hadzliana Zainal, Kok-Lim Alvin Yau and Soo-Huat Teoh in DIGITAL HEALTH

Supplemental Material

sj-pdf-2-dhj-10.1177_20552076251329991 - Supplemental material for Healthcare professionals’ perspectives on artificial intelligence (AI)-based mobile applications (apps) for diabetes education and behavioural management

Supplemental material, sj-pdf-2-dhj-10.1177_20552076251329991 for Healthcare professionals’ perspectives on artificial intelligence (AI)-based mobile applications (apps) for diabetes education and behavioural management by Phei-Ching Lim, Yung-Wey Chong, Qiu-Ting Chie, Hadzliana Zainal, Kok-Lim Alvin Yau and Soo-Huat Teoh in DIGITAL HEALTH

Footnotes

Acknowledgements

This work was supported by the Fundamental Research Grant Scheme (FRGS/1/2023/ICT02/USM/02/3), Ministry of Higher Education, Malaysia. We thank the Director General of Health, Ministry of Health Malaysia for the permission to publish this article.

ORCID iD

Yung-Wey Chong

Ethical considerations

This research is registered in the National Medical Research Register, Malaysia (No. NMRR-23-01186-APO) and approved by the Medical Research Ethics Committee, Ministry of Health, Malaysia. It was also registered with Universiti Tunku Abdul Rahman’s Scientific and Ethical Review Committee (U/SERC/27/2023).

Informed consent

All participants received an information sheet and written informed consent was obtained. All research procedures were conducted according to ICH-GCP and other regulatory requirements.

Author contributions/CRediT

All authors contributed equally to the conception, design, preparation of the study, data collection, analysis, writing, and review of the manuscript. All authors read and approved the final manuscript.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The research was supported by the MOHE, Malaysia through the Fundamental Research Grant Scheme FRGS/1/2023/ICT02/USM/02/3.

Conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Supplemental material

Supplemental material for this article is available online.

References

International Diabetes Federation. IDF diabetes Atlas 2021. 10th ed. [Internet], https://diabetesatlas.org/atlas/tenth-edition/ (2021, accessed 15 May 2024).

National Institute of Public Health. National health and morbidity survey 2019 non-communicable diseases, healthcare demand, and health literacy – key findings. 2019.

Yau

KLA

Chong

Fan

, et al. Reinforcement learning models and algorithms for diabetes management. IEEE Access 2023; 11: 28391–28415.

Logan

. Transforming hypertension management using mobile health technology for telemonitoring and self-care support. Can J Cardiol 2013; 29: 579–585.

Long

Yang

, et al. The effect of smartphone app-based interventions for patients with hypertension: systematic review and meta-analysis. JMIR Mhealth Uhealth 2020; 8: e21759.

Padwal

Wood

. Digital health approaches for the assessment and optimisation of hypertension care provision. Can J Cardiol 2021; 37: 711–721.

Varnfield

Karunanithi

Lee

, et al. Smartphone-based home care model improved use of cardiac rehabilitation in postmyocardial infarction patients: results from a randomised controlled trial. Heart 2014; 100: 1770–1779.

Song

Ren

Liu

, et al. Effect of smartphone-based telemonitored exercise rehabilitation among patients with coronary heart disease. J Cardiovasc Transl Res 2020; 13: 659–667.

Yudi

Clark

Tsang

, et al. SMARTphone-based, early cardiac REHABilitation in patients with acute coronary syndromes: a randomized controlled trial. Coron Artery Dis 2021; 32: 432–440.

10.

Díaz-Mohedo

Carrillo-León

Calvache-Mateo

, et al. App-Mohedo®: a mobile app for the management of chronic pelvic pain. A design and development study. Int J Med Inform 2024; 186: 105410.

11.

Lee

Wong

Lee

SWH

. m-Health views and perception among Malaysian: findings from a survey among individuals living in Selangor. Mhealth 2020; 6: 6.

12.

IXY

Kee

JCY

Threapleton

, et al. Effectiveness of smartphone technologies on glycaemic control in patients with type 2 diabetes: systematic review with meta-analysis of 17 trials. Obes Rev 2018; 19: 825–838.

13.

Joachim

Forkan

ARM

Jayaraman

, et al. A nudge-inspired AI-driven health platform for self-management of diabetes. Sensors 2022; 22: 4620.

14.

Liu

Lin

, et al. Implementing a novel machine learning system for nutrition education in diabetes mellitus nutritional clinic: predicting 1-year blood glucose control. Bioengineering 2023; 10: 1139.

15.

Kwan

Cheng

Yoon

, et al. A systematic review of nudge theories and strategies used to influence adult health behaviour and outcome in diabetes management. Diabetes Metab 2020; 46: 450–460.

16.

Michiura

Matsuda

Suwa

, et al. A self-management system for preventing hyperglycemia through blood glucose level prediction and nudge-based food amount reduction. In: 2023 45th annual international conference of the IEEE Engineering in Medicine and Biology Society (EMBC), 24–27 July 2023, Sydney, Australia, pp.1–7. IEEE.

17.

Abdullah

Liew

, et al. Health literacy experiences of multi-ethnic patients and their health-care providers in the management of Type 2 diabetes in Malaysia: a qualitative study. Health Expect 2020; 23: 1166–1176.

18.

Steinman

Heang

van Pelt

, et al. Facilitators and barriers to chronic disease self-management and mobile health interventions for people living with diabetes and hypertension in Cambodia: qualitative study. JMIR Mhealth Uhealth 2020; 8: e13536.

19.

Thies

Anderson

Cramer

, et al. Lack of adoption of a mobile app to support patient self-management of diabetes and hypertension in a federally qualified health center: interview analysis of staff and patients in a failed randomized trial. JMIR Hum Factors 2017; 4: e7709.

20.

Al-Qazaz

Hassali

Shafie

, et al. Perception and knowledge of patients with Type 2 diabetes in Malaysia about their disease and medication: a qualitative study. Res Social Admin Pharm 2011; 7: 180–191.

21.

Bonet Olivencia

Rao

Smith

, et al. Eliciting requirements for a diabetes self-management application for underserved populations: a multi-stakeholder analysis. Int J Environ Res Public Health 2021; 19: 127.

22.

Kong

Moy

Ong

, et al. MyDietCam: development and usability study of a food recognition integrated dietary monitoring smartphone application. Digit Health 2023; 9: 1–18.

23.

Guest

Bunce

Johnson

. How many interviews are enough? An experiment with data saturation and variability. Field Methods 2006; 18: 59–82.

24.

Saunders

Sim

Kingstone

, et al. Saturation in qualitative research: exploring its conceptualization and operationalization. Qual Quant 2018; 52: 1893–1907.

25.

Baker

Edwards

. How many qualitative interviews is enough? Expert voices and early career reflections on sampling and cases in qualitative research. In National centre for research methods review paper, National Centre for Research Methods, Economic and Social Council (ESRC), United Kingdom, 2012.

26.

Dworkin

. Sample size policy for qualitative studies using in-depth interviews. Arch Sex Behav 2012; 41: 1319–1320.

27.

Ibrahim

Rashad

. The evolving role of healthcare professionals in the age of AI: impacts on employment, skill requirements, and professional development. J Artif Intell Mach Learn Manage 2024; 8: 14–21.

28.

Esmaeilzadeh

. Use of AI-based tools for healthcare purposes: a survey study from consumers’ perspectives. BMC Med Inform Decis Mak 2020; 20: 1–19.

29.

Yousif

Asghar

Akbar

, et al. Exploring the perspectives of healthcare professionals regarding artificial intelligence; acceptance and challenges. BMC Health Serv Res 2024; 24: 1–9.

30.

Shinners

Grace

Smith

, et al. Exploring healthcare professionals’ perceptions of artificial intelligence: piloting the shinners artificial intelligence perception tool. Digit Health 2022; 8: 20552076221078110.

31.

Hua

Petrina

Young

, et al. Understanding the factors influencing acceptability of AI in medical imaging domains among healthcare professionals: a scoping review. Artif Intell Med 2023; 147: 102698.

32.

Aquino

Rogers

Braunack-Mayer

, et al. Utopia versus dystopia: professional perspectives on the impact of healthcare artificial intelligence on clinical roles and skills. Int J Med Inform 2023; 169: 104903.

33.

Sonawane

Shah

Pote

, et al. The application of artificial intelligence: perceptions from healthcare professionals. Health Technol (Berl) 2023; 13: 861–867.

34.

Lynn

. Determination and quantification of content validity. Nurs Res 1986; 35: 382–386.

35.

Davis

. Instrument review: getting the most from a panel of experts. Appl Nurs Res 1992; 5: 194–197.

36.

Clarke

Braun

. Teaching thematic analysis: overcoming challenges and developing strategies for effective learning. Psychologist 2013; 26: 120–123.

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.

0.00 MB

0.03 MB

0.08 MB