We included focus groups based on the Consolidated Framework for Implementation Research. All focus group sessions were audio recorded and transcribed by a professional service, Atomic Scribe. Subsequently, the focus group interview transcripts were coded and categorized into themes used for affinity mapping.

The research team (JF, TC, BA, MK, AF, TM, and KN) and members of a Community Advisory Board actively recruited potential participants from the New Haven community. We posted flyers throughout the New Haven community including in the Yale Eye Center, the Dana Eye Center, Yale affiliated primary care physician (PCP) and endocrinologist’s offices, recreational centers, community restaurants, churches, and recruited by word-of-mouth. We also reached out to individuals who previously expressed interest in participating in studies at the Sight-Saving Engagement and Evaluation in New Haven (SEEN) Lab. For inclusion in the study, individuals needed to (1) be at least 18 years old, (2) have a diagnosis of diabetes, (3) speak English fluently, and (4) have reliable transportation to attend the in-person focus group.

We screened potential participants via telephone using a screening questionnaire. Each focus group session lasted approximately 90 minutes and included a dinner that accommodated participant food allergies and diet preferences.

The institutional review board (IRB) at Yale School of Medicine approved the study and all participants gave informed consent (IRB #2000034710). This study adhered to the Declaration of Helsinki. Participants were consented before the beginning of any research activities, including the baseline assessment, by the research assistants (JF, TC, AF, BA, and MK), who were trained in the process. The Yale IRB reviewed all materials for ethical standards, and compliance. Privacy and confidentiality standards were maintained through the use of deidentified data for sharing with data analysts, and Protected Health Information was kept digitally on a private, encrypted server, while paper records containing Protected Health Information were kept in a secure, locked area that only provided access to the research team. Participants who completed the study were each compensated with a gift card containing $50 US dollars.

The research team developed an interview guide Multimedia Appendix 1. We developed the interview guide using a community-based approach by consulting with an advisory board comprised of members from the community. We collaborated with members of our Community Advisory Board to refine the questions, ensuring they accurately addressed specific community needs and contexts (Multimedia Appendix 2). The focus group interview guide included questions about the following topics: (1) participants’ experiences with diabetes (2) their eye health awareness, (3) their technology usage, and (4) their preferences for features in a digital tool to learn about eye health and screening. The team refined the interview guide during data collection to include emergent themes and improve question clarity. This iterative process allowed the team to adapt to participant feedback and incorporate prompting questions to elicit comprehensive responses.

We conducted semistructured focus group interviews in 2 community spaces. Results from a study by Guest et al [27] revealed that more than 80% of all themes are discoverable within two to three focus groups, and 90% of themes could be discovered within 3 to 6 focus groups. Therefore, 4 focus group sessions were conducted. Each session was capped at 6 participants to allow for all participants to fully engage in the discussions. We ensured the setting was conducive to discussion, collaboration, and confidentiality. AF (medical student) moderated each session, and multiple notetakers (VR, KN, TM, JF, TC, MK, and BA) attended different sessions to take notes on body language, nonverbal cues, and emerging themes. The moderator and notetakers debriefed after each interview. For the first 30 minutes of the focus group session, notetakers and the moderator ensured that participants completed the demographic surveys and signed the consent forms after demonstrating full understanding of them. We informed participants that all information would be deidentified and used solely for study purposes. The sessions were recorded with an audio recorder and we used a backup recorder on the computer. At the end of the sessions, each participant was compensated with a US $50 gift card. After each focus group, the moderator and notetaker debriefed. We sent all audio recordings to Atomic Scribe for transcription and redaction of identifying information to ensure participant confidentiality. We securely stored all data in an encrypted server.

The analytic team was comprised of members from the SEEN Lab (AF, TC, JF, KN, VH, AG, and SA) and the Consumer Health Informatics Lab (TM, BA, MK, AL). Team members included a vitreoretinal surgeon, retina fellow, consumer health informatics expert, user experience and user interface designer, medical students, graduate students with advanced degrees in business, health sciences, and public health and researchers with extensive experiences working with marginalized populations.

Thematic saturation was achieved with 4 focus groups (n=19, 2‐6 participants each focus group session). We recruited 24 individuals, of whom 19 attended one focus group session. Focus group sessions were approximately 90 minutes long (60 min devoted to the semistructured interviews). The study participants included a diverse demographic, with 68% (13/19) female, 47% (9/19) identifying as Black, 26% (5/19) as Hispanic, and 11% (2/19) as indigenous. Over 80% (15/19) of participants had access to smart devices, defined as either smartphones, smartwatches, computers, and tablets.

Most participants, over 70% (13/19), disclosed annual household incomes below US $50,000. Hemoglobin A_1c (HbA_1c) was self-reported (mean 6.77, SD 1.93). Ages ranged from 35 years old to 72 years old (mean age 58.79, SD 19.54). Approximately half (10/19, 52%) of participants did not pursue higher education (neither partially nor completed a college degree). Over one quarter (5/19, >26%) of participants reported having a diagnosis of DR. Of note, one group consisted of all women, and another group included 3 participants with physical disabilities.

During the initial rapid qualitative analysis, the research team (KN, BA, AF, MK, TC, and VH) presented the main takeaways from the focus group sessions. The main findings highlighted the app should include: (1) education and resources on diabetes, (2) ability to communicate with providers, (3) options to personalize the user experience, (4) features that facilitate peer support, and (5) components that address emotional and mental stress associated with diabetes diagnosis and management. Supporting quotes for the main themes are presented in Table 1.

Participants expressed a desire for diabetes-specific diet advice, particularly regarding specific food choices and how much to eat as well as finding ways to exercise. Some participants highlighted a need for easy recipes and meal plans for cooking at home, incorporating nutrition information and comprehensive recommendations. When dining out, participants said they would benefit from education regarding what is considered fast food, and what take-out options are better choices. Participants expressed frustration at the obscurity of food labels and wanted education on reading them to better understand their food’s nutritional content. Participants understood the importance of exercising but found it difficult to exercise, especially when it came to feeling motivated. Participants expressed wanting resources for all types of skill levels in exercising and having a group event to motivate them to exercise.

Participants sought to know what the effects of diabetes are on their health. Many participants desired informative content in the app regarding diabetes. For example, participants requested lists of “do’s and don’ts for people with diabetes” and “frequently asked questions.” Some participants also requested comprehensive information regarding the complications and side effects of diabetes, results from new diabetes-related research, and information about novel treatments and their effects.

When it came to understanding how diabetes affects the eyes, some participants shared that they were informed about the risk of vision problems and the importance of eye care at the time of their diagnosis. In contrast, others expressed that they were not told that diabetes can lead to blindness. Some learned about eye care through personal experiences or from family and friends, while others were informed by their health care providers. Many recognized the benefits of regular eye exams, with most having had an exam within the past year to a few months ago, despite some difficulty scheduling appointments. The group had a positive attitude toward diabetes eye care, although experiences varied. Participants also expressed a desire to better understand how diabetes affects the eyes.

Participants spoke of the barriers that they face accessing health care. Long wait times to scheduling an appointment with a physician was a common topic, as was limited access to health care. Participants also reported challenges with the sheer amount of medications to take and their associated cost, availability, and side effects. Several participants reported financial barriers, citing the high overall cost of care. Many participants were well-versed in the need to make lifestyle changes but reported barriers to implementing them including transportation, the increasing cost of healthy foods and comparatively high accessibility of fast food.

Participants expressed a desire for a user-friendly digital health tool that is informative without being overwhelming. They suggested features such as a point reward system, customization options, and streamlined content. A key priority was improving communication with healthcare providers, with participants highlighting the need for supplementary materials and guidance to facilitate conversations with doctors. They also showed interest in opportunities to participate in research studies and clinical trials. Some participants spoke about having a digital tool that makes records easy to access, like MyChart.

Participants discussed their preferences for receiving information, including push notifications from the app. They valued notifications about new treatments, support groups, upcoming doctor’s appointments, medication reminders, exercise reminders, blood glucose, and blood pressure tracking.

Participants consistently identified family as a source of support. Some participants had older family members with diabetes, so they grew up witnessing the experience of diabetes, forming a unique perspective. Others spoke of friends and community as a resource. Support for these individuals came from community health centers and health care providers. In contrast, other participants described a need for stronger support due to lack of knowledge or commitment from their existing family or community. Participants voiced that they would like to have a platform to build community.

Participants spoke of the emotional toll that their chronic disease has had on them. They described feelings of depression, denial, embarrassment, and psychosocial struggles. They spoke of the emotional journey that living with diabetes engenders and seeking answers to how diabetes complications can affect their mental capacity.

Figure 1 presents features organized by impact potential and implementation difficulty. The development of the digital health tool prioritized the must-have features, located to the upper left quadrant. These were considered the most important to the participants and feasible for the research team to execute. The digital health tool will include reliable content about eye care, diabetes, and related health topics, addressing the need for trustworthy information about diagnosis, medications, and treatments. To further promote holistic wellness, it will offer resources for exercise, cooking videos, mental health, and general wellness, and other functions to provide, promote, and support overall well-being. Features will inform users of local events, particularly those related to diabetes, offer resources for clinical trials, established programs, and integrate relevant existing applications for managing patient health data. The advisory board showed consensus on the must-have features. Additional recommendations included adding testimonials and tutorials to the “Frequently Asked Questions” section, and establishing partnerships with local businesses, advocacy organizations, and public works departments, such as the city health department and the National Association for the Advancement of Colored People, to address social needs. In addition, the advisory board highlighted the value of community groups and discussion forums for individuals with diabetes.

Our findings highlight the need for a culturally responsive digital health tool for individuals with diabetes. This tool will incorporate lifestyle resources, informal support platforms, and electronic health record access while maintaining a well-calibrated notification system. These findings align with previous research on developing smartphone apps for diabetes management, emphasizing the importance of usability, patient-centered resources, and seamless integration with existing health care infrastructures. The consistency with other studies highlights the critical role of tailored digital tools in enhancing diabetes care and patient engagement [9,10,12]. The need for a patient-centered digital health tool that provides access to resources like guided communication with health care providers, eye health, and diabetes education. Key features include personalized user experiences, customizable notifications, peer support, and community-building platforms. Consistent with existing literature, our study confirms that patients with chronic conditions like diabetes prefer technology that supports online community engagement and self-management resources, including exercise routines, healthy recipes, and telehealth options, to enhance their overall health management [38].

Studies show that technological innovations for diabetes management, particularly through mobile devices, can improve health outcomes [10,39]. Low-cost, user-friendly digital tools that centralize information and functionality can facilitate patient autonomy and improve access to health guidance. These technologies can further inform medication management, blood glucose monitoring and subsequent communication with health care professionals [40]. A recent benchmark study evaluated the usability and content of available mobile apps for people with diabetes [41]. In their systematic evaluation, most apps had good usability and potential to facilitate diabetes management. Common app features included nutritional and dietary planning, monitoring glycemic control, and scheduling health appointments. Content areas that were less well represented included mental health, education on diabetes complications, and content aimed at individuals with disabilities. No apps existed with content related to DR or resulting visual impairment [41]. While many digital tools do exist, there are significant content gaps among available resources. There is a widespread need for an app that provides reliable information and guidance for people with diabetes who want to prevent or manage their visual impairment resulting from DR.

In addition to content within an app, the design of an app impacts its success. Specifically, the screen or display size, color, material, menu structure, user-friendliness, accessibility, and simplified operations can influence user adoption and use, particularly among older adults [42,43]. Therefore, when designing interventions or applications designed for older adults with diabetes, health care providers or app developers should focus on aesthetic features that consider the physical, sensory, and cognitive changes experienced by older adults, such as impaired vision, slow hand movement, and decreased response time [44-46].

The impact of chronic disease on mental health is well-documented. Individuals living with type 1 or type 2 diabetes are at heightened risk for depression, anxiety, and eating disorders [47,48]. Many people with chronic diseases, including diabetes, use the internet to access psychological treatments [49-51]. Baldwin et al [52] explored the use of the web-based electronic mental health program, myCompass. MyCompass is both clinically effective and cost-efficient in expanding access to mental health services [50] and has previously demonstrated efficacy in reducing depressive symptoms and diabetes-related distress in individuals with type 2 diabetes. However, a notable limitation of MyCompass is that being primarily a mental health support program, it does not provide support targeting specific diabetes needs [52]. More recent studies incorporate diabetes-specific content [53] alongside existing depression treatments and cognitive behavior therapy [54]. Participants reported significant improvements in health behaviors, such as healthier eating habits, better health monitoring, and incremental progress in overall outcomes [52]. With increasing recognition of the importance of mental health in diabetes care [55], our study acknowledges the need to address mental health in the management of diabetes in the design of our proposed digital health tool.

Several studies emphasize the critical role of social support in diabetes management [56-58]. Toobert and Glasgow [59] identified it as a vital factor in self-care adherence, while Marquez et al [60] noted its impact on physical activity and weight loss. Shayeghian et al [61] linked social support to better blood glucose control, and Pereira et al [62] highlighted its importance for self-care behaviors, such as glucose monitoring. Diabetes online communities have emerged as valuable platforms for peer support, fostering communication, improving disease management, and enhancing psychosocial well-being [63].

There are few studies on the role of online peer support in diabetes care. One study used digital observation, applying ethnographic techniques to study interactions on social media and the web to analyze hundreds of diabetes-related posts on platforms like Facebook (Meta), Twitter (X), and YouTube (Google), identifying 6 core themes—humor, diabetic pride, technology use, tips and tricks, community building, and venting [64]. The findings emphasize the importance of online platforms in fostering camaraderie, peer mentorship, and shared experiences often overlooked by health are systems. In addition, it has been shown that integrating mental health services and peer support is beneficial and critical for positive short- and long-term diabetes outcomes [64-66]. Our digital tool aims to replicate these positive outcomes by creating a supportive environment for patients to connect with each other and share their experiences, thereby enhancing their overall care.

A significant concern for participants was their lack of knowledge about how diabetes could affect them. These gaps included understanding the full range of symptoms, complications, predisposing factors, and psychological impacts of diabetes. Research shows that individuals without diabetes self-management education are less likely to engage in preventative care, leading to worse outcomes [67]. Limited health literacy is recognized as a significant barrier to improving health outcomes. Individuals with low health literacy may have less knowledge about diabetes, have less control of blood glucose levels, engage in less self-management behaviors, and incur higher health care costs [68]. One study demonstrated a direct correlation between health literacy and several key factors in diabetes management, including self-monitoring, nutrition and exercise habits, diabetes knowledge, self-care practices, and social support [69]. Another study recommended that targeted educational and behavioral interventions are crucial in addition to health care service strategies, particularly for older adults and individuals with less formal education [10,65,66].

With the rise of social media use, patients with diabetes are more likely to seek information using technology to help manage their condition [70]. In one study, nearly half of participants sought self-management information online, with dietary planning being the most common focus [71,72]. The term eHealth literacy refers to the critical skills needed to effectively use technology for seeking, accessing, and understanding health information from electronic sources [73,74]. Previous research demonstrates a significant association between eHealth literacy and health behaviors. Lower eHealth literacy is linked to poor diabetes self-management and worse health outcomes, particularly among older adults with diabetes [75,76].

Incorporating design preferences of intended users into digital health tools may increase participant engagement and empower them to manage their health better [13,14]. This includes trust-centered design elements such as reliable content on eye care, diabetes, social support, and healthy lifestyles [13,14]. Studies suggest that digital tools should be specifically designed to address diabetes-related concerns [58,77,78].

These findings align with the discussed topics that participants acknowledged and the importance of incorporating eHealth literacy and social support into digital health tools.

Some existing apps related to DR include the National Eye Institute’s virtual reality app, which simulates vision loss, and the RetinaRisk (Retina Risk) app, which calculates personalized risk for diabetic retinopathy. However, few apps focus extensively on DR content or address visual impairment resulting from the disease. Our participants expressed a strong desire for this type of content. Our digital health tool will fulfill this gap by incorporating trusted resources for education, research opportunities, community events, and mental health support tailored to eye health and diabetes care.

Although this study has many strengths and advantages, there are several limitations and disadvantages when conducting focus group discussions as part of the methodology. Some challenges include limited control over data collection, the influence of group dynamics on participants’ responses, and the difficulty in generalizing findings to a larger population.

Focus groups can sometimes include silent or dominant participants, which may influence the conversation dynamics and impact data collection [79]. To address this, the moderator (AF) and observers (VH, KN, TM, JF, TC, MK, and BA) implemented strategies to ensure that all participants had an opportunity to share their views, such as redirecting the discussion to stay on topic, rephrasing questions to allow wider range of responses, and seeking clarification to encourage the expression of diverse viewpoints and perspectives.

Our focus groups had homogeneous backgrounds (diabetes status and geographic location), while also having heterogeneous traits (age differences, gender, race, and ethnicity) that ensured a diversity of perspectives and greater depth of analysis into the challenges they faced [80]. Although participants in this study shared similar opinions and perspectives, their experiences may not be generalizable to the broader population of individuals with diabetes mellitus, particularly given the geographical restriction of recruitment of the participants, which limits the national representativeness of the sample [81].

Participant recruitment included the participants self-reporting their health measures such as current hemoglobin A_1c. This form of providing information can introduce a potential source of error or bias due to self-reporting that cannot be verified [82]. The SEEN tool has the potential to address this limitation by allowing participants to input their information or data more accurately for recall and reporting. For instance, the tool may be interoperable with glucose monitors, allowing participants to integrate their glucose data into the tool, thereby minimizing the likelihood of self-report errors.

This study did not collect household size data, which limits our ability to infer whether individuals met the criteria for poverty. However, the experience of multiple individuals manifesting through direct quotes from them is that medication costs, food costs, and other financial barriers impact access to eye care and screening. Whether or not any participants have experienced financial difficulty with managing their diabetes, their statements point to the universality of financial barriers to health care equity.

Despite these limitations, the study results yielded key insights that will inform the design team on what community needs to prioritize in the development of the digital health tool. Furthermore, our community-based approach lifts the voices of participants who are historically under-represented in medical research, and empowers them with lasting educational benefits and other resources to facilitate their management of DR.

The main findings of this study serve as a guide for the design priorities for a digital health tool tailored to individuals with diabetes or DR.

The next steps, which is currently underway, will be to develop an initial prototype of the digital health tool through 3 phases of production: design, development, and evaluation of the digital health tool. During the design phase, we will create detailed user journey maps, develop health education content for the tool, and develop low-fidelity wireframes and interactive high-fidelity prototypes. All of this will be worked in collaboration with the community members to incorporate feedback and ensure the tool is intuitive and user-friendly. After the design phase, the team will begin the development phase to build the digital health tool by collaborating with developers to build the tool. Finally, the final phase will be to conduct usability testing and implement a pilot study to collect quantitative and qualitative data to test the prototype with application users in order to ensure the tool meets the community’s needs.

Future studies can expand the scope and the focus groups. The focus groups can consist of more diverse populations, including different age groups, genders, socioeconomic statuses, and cultural backgrounds, to gather a wide range of perspectives. Researchers could also include participants from various geographic locations as well to understand regional differences in digital health needs and preferences.

Using a user-centered design approach, when developing digital health tools, has the potential to significantly improve adverse health outcomes for individuals living with diabetes through culturally relevant and personalized interventions. The community-centered approach not only empowers participants from communities which are often unheard to give their voice to health care research, but will also educate them through an educational component, including connecting the participants to resources to help them manage DR.