This study used descriptive qualitative research design, which followed the principles of naturalistic inquiry [21]. Face-to-face semistructured interviews with the older adults were conducted between November 2024 and December 2024.

Participants were recruited from urban parks and leisure clubs for older adults in Beijing, China, where they engaged in regular recreational and social activities. These sites were chosen because they are popular gathering places for community-dwelling older adults, making them ideal for reaching individuals with diverse levels of ICs. However, we acknowledged that such sampling might introduce a selection bias toward relatively active older adults who are more socially involved. Purposive sampling was used in this study, which enabled researchers to select experienced participants based on the aims of the study [22]. The maximum variation sampling technique was also applied simultaneously to recruit heterogeneous samples of older adults with different levels of ICs, education, sex, and age [23]. The participants had different levels of ICs which were measured by traditional assessment methods (Table S1 in Multimedia Appendix 1), and the educational attainment of participants included primary, secondary, and higher education, to ensure diversity in cognitive and experiential perspectives, offering a wealth of diverse perspectives that could enrich the research topic.

Potential participants were selected according to the following criteria: (1) aged 60 years or older, (2) completed the IC assessment (Table S1 in Multimedia Appendix 1), (3) able to communicate effectively in daily life, and (4) informed consent and voluntary participation in this study.

Older adults were invited to participate in both the assessments of ICs and qualitative interviews. The IC assessment was conducted using a tablet computer, with specific measurement tools and scoring rules detailed in Table S1 in Multimedia Appendix 1. Interviews were scheduled with individuals who agreed to participate after completing the IC assessment. All participants were assured that their participation was entirely voluntary, and written informed consent was obtained prior to data collection.

Before the interview, the interviewer fully introduced the objective of this study to the older adults. The researcher prepared textual and graphic materials to explain the study to the older adults, helping them understand their ICs and DT technology. In addition, a short video introducing DT technology and the concept of a digital avatar was shown to enhance participants’ comprehension. Everyday examples and scenario-based demonstrations were used to ensure comprehension, and participants were asked to restate their understanding before the discussion began. The older adults were encouraged to actively share their understanding of the applications of the avatar and DT technology, as well as to voice their perspectives and related knowledge during the discussion. Afterward, the older adults were invited to ask any questions they had about the avatar or DT technology, and the researchers provided clear and comprehensive responses to their inquiries. The participants were also told they could withdraw at any point during the interview. The interviews were conducted in a quiet and relaxed cafe environment that allowed for open and comfortable conversation. The interviews were conducted through Chinese conversations. The participants were encouraged to express themselves based on the interview guide. Audio recordings were made with the consent of the participants.

All of the interviews were conducted by the same researcher (YN) to ensure consistency. The researcher was trained by qualitative research experts and practiced interviews to refine their interviewing skills. The interviews were conducted using a semistructured interview guide (Table S2 in Multimedia Appendix 2), which was initially designed based on self-construal theory, providing a deeper understanding of individuals’ self-perception [24]. Then, experts in qualitative research, IC, and DT technology were organized to discuss the interview guide, and the interview guide was revised based on feedback. One participant, who met the inclusion criteria, was selected for a preinterview, and the interview outline was modified based on the results of this preinterview. The data obtained from the preinterview were not included in the analysis. The principle of data saturation was a recognized criterion for determining sample size in qualitative research designs [22]. In this study, data collection and analysis processes were carried out simultaneously, and data collection was stopped when no new data was generated related to the topic of study. Specifically, after interviewing 21 participants, no additional themes related to the study topic were identified, and 2 further interviews were conducted to confirm saturation. Therefore, data collection was stopped once thematic redundancy was observed.

The interview recordings were transcribed and organized verbatim within 24 hours of the end of each interview. NVivo 12.0 software (QSR International Pty Ltd) was used for coding and analysis, and data analysis was conducted in Chinese. This study adopted the content analysis approach [25]. The interviewer reviewed the text multiple times to ensure comprehensive understanding of its content. Subsequently, open coding was conducted to identify key ideas and concepts within the material. To ensure comprehensive capture of information, open coding was conducted in a line-by-line, sentence-by-sentence format. Then, similar and related codes were classified into categories and subcategories. Furthermore, themes and subthemes were generated based on categories and subcategories. This process was iteratively repeated until saturation was reached, with no new themes or subthemes emerging, and was independently conducted, confirmed, and discussed by 2 researchers. Finally, the themes and subthemes were translated into English using both direct translation and back-translation methods.

Rigor focuses on 4 key dimensions: credibility, transferability, dependability, and confirmability [26]. The credibility was strengthened by the researcher’s establishment of trusting relationships with older adults, which facilitated authentic self-expression during the interview process. The results of the study were also fed back to the participants, enabling them to confirm and add to the results. Additionally, the researcher used reflexive journaling to track the research process and identify potential biases, thereby enhancing credibility. To enhance transferability, a purposive sampling method was adopted to ensure that participants had rich perspectives. Moreover, rich interview data enhanced transferability by documenting participants’ context, enabling evaluation of results’ applicability to other settings. The dependability was addressed through regular discussions among the researchers. These discussions allowed for the resolution of disagreements and confusion during the analysis process. Finally, the confirmability was established by providing a description and rationale for the decisions made based on the results.

This study was approved by the ethics committee of the Capital Medical University (batch: Z2023SY138). Informed consent was obtained from all participants, and the study adhered to all relevant privacy and confidentiality protocols. All patient-related information has been fully anonymized and de-identified. No compensation was provided to the participants. All procedures were conducted in accordance with the guidelines of the Institutional Review Board (IRB) and Research Ethics Board (REB).

A total of 23 older adults were interviewed face-to-face, with participant selection considering the age, sex, degree of education, and the level of IC. Interviews lasted between 27.75 and 49.58 minutes (mean 37.57, SD 6.77 min). The demographics were outlined in Table 1. Of the 23 older adults, 12 were women and 11 were men, with an average age of 68.30 (SD 5.45; range: 61-84) years. 17 had a high school education or higher, including 6 with a bachelor’s degree or higher. The average score for IC was 4.04 (SD 0.98; range: 2-5).

The appearance of the digital avatar served as the end point for the presentation to older adults and directly affected the effectiveness and adherence of the application. According to the concept of DT technology, the digital avatars were replicas of physical entities that synchronized with the changes in real older adults. Unexpectedly, they voiced some perspectives that were outside the box.

DT technology can map the ICs of older adults, and the shape and changes of the digital avatar indicate the level and changes in their ICs. The expectations among older adults regarding how DT should reflect the diversity and dynamics of their ICs vary widely.

From the perspective of older adults, DT offers the following potential benefits.

Older adults expressed some expectations for their DTs, notably the following subthemes.

Based on the insights, we further developed a conceptual model, the “windmill” model, to better understand how to build DTs of older adults to map their ICs (Figure 1). Centered on the evaluation of IC, the “windmill” model integrated DT technology and comprised 4 key components. First, the appearance of the DT, which reflects 2 main preferences: “mirror-like” and “idealized” forms. Second, the mechanism of mapping IC, which enables dynamic visualization of older adults’ IC levels and changes through multimodal presentation and behavioral mapping. Third, feedback service, which includes 4 functional roles: health advisor, life assistant, spiritual mentor, and role supplement. Fourth, interactive expectations, which emphasize intelligence and knowledge integration, the provision of positive feedback, effective coordination among families, communities, and health care institutions, and ensuring the security of personal privacy information. Grounded in the perspective of older adults, this “windmill” model provided support for the application of DT technology in IC assessment.

The integration of DT technology into IC assessment held the potential to transform how older adults’ IC was monitored and supported. Incorporating the perspectives of older adults, this study offered an initial exploration of how DT technology might be leveraged to support IC assessment. As for avatar appearance, the participants have demonstrated diverse perspectives regarding their envisioned characteristics of an ideal DT. They also described how the DT should represent and map their IC levels, highlighting the importance of multimodal reminders for enhancing IC comprehension. In addition, participants expected the DT to offer feedback services that could help maintain or improve IC. Based on these findings, a preliminary framework—the “windmill” model—was developed, incorporating 4 core components: appearance, intrinsic capacity, feedback services, and interactive expectations. The “windmill” model offered potential pathways for guiding the design of DT-supported IC assessment tools for older adults.

The appearance of digital avatars played a pivotal role in user engagement and acceptance, which was conducive to helping older adults understand and maintain their ICs. The findings of this study emphasized the importance of personalized digital avatars, consistent with existing literature that indicated superior self-identity recognition with personalized digital representations versus generic avatars [27]. The digital avatars might be one-to-one reductions of physical entities. Individuals might demonstrate an inherent desire to protect avatars that mirrored themselves [28,29], which was beneficial for them to maintain their internal abilities. Digital avatars representing older adults were also expected to take other forms, such as younger versions of themselves, specific organs, or even their idols. It has been shown that whether or not digital avatars resemble the individuals themselves, both can have an impact [30]. Effective interaction, rather than mere personification, might be the key to the success of digital avatars [31]. Therefore, the design of digital avatars should be tailored to the unique characteristics and preferences of older adults to enable effective interactions, rather than strictly replicating their physical appearance.

DT technology realized the dynamic mapping of the ICs of older adults. DT technology could effectively reflect ICs in the following ways: (1) current IC levels and trends, presented through multimodal methods such as voice, text, images, and animations, which cater to diverse user preferences and cognitive abilities; and (2) synchronous changes between digital avatars and the older adults they represent, which were more easily understood. By bridging the gap between complex health data and actionable information, DT technology had the potential to empower older adults to better understand and manage their ICs, which could compensate for the health inequalities caused by the digital divide [32].

The digital avatar had a dual identity—“from avatars to agents” [27]. In this study, digital avatars functioned not only as visual representations to display the ICs of older adults but also as interactive agents capable of providing feedback services. These feedback services focused on health support, life support, and psychological and social support. In terms of health support, digital avatars could provide health status monitoring, early warning, and health education [14]. Additionally, when interacting with digital avatars representing themselves, individuals could adopt an interventionist perspective, leading to a more objective self-assessment [33]. Digital avatars were also expected to provide assistance in daily life for older adults, thereby increasing user engagement and stickiness. Regarding emotional and social support, older adults could identify with their digital avatars, experience meaningful emotions, and form authentic attachments with them. Digital avatars created a beneficial psychological distance that facilitated more objective self-reflection while maintaining emotional engagement [34]. Compared to interactions with real people, individuals perceived digital avatars as more private and secure, making them more willing to express themselves [34,35]. Avatar therapy has been used for the treatment of psychosocial diseases [35-37]. In summary, digital avatars were expected to serve as multifunctional tools.

Older adults expressed some expectations for DT technology in mapping their ICs. First, the process should be intelligent and equipped with sufficient knowledge. In addition to the foundational technologies such as sensors, Internet of Things, machine learning, and deep learning, the intelligence of DT can be further advanced through the integration of multisensor fusion, edge-cloud computing synergy, virtual reality or augmented reality interactive technologies, and generative adversarial networks for data augmentation [38,39]. In the future, DT could be further enhanced by integrating knowledge graphs [40] and expert knowledge [11] to ensure sufficient knowledge support. Second, older adults need positive feedback from DT. Positive feedback has been proven to improve self-efficacy and intrinsic motivation [41]. Third, the DT technology, as a multisubjective bond, connected the older adults, families, communities, and hospitals. From the perspective of older adults, remote health care services become more accessible. For families, it provides a clearer and more comprehensive understanding of older adults’ well-being. Meanwhile, communities and hospitals can achieve batch management of older adults. Finally, the privacy and security of DT technology can be guaranteed by blockchain technology, federated learning, and other methods [11].

Overall, the “windmill” model provided insights from older adults on the construction of the virtual entity, the mapping of IC, the mechanisms of interaction between the virtual entity and his or her twin, and the types of services that a DT system should deliver. It also encompasses the core components of a DT system—the physical entity, its virtual twin, and the mapping relationship linking the two [42]. Compared with the 5D DT model, the “windmill” model offered a distinct theoretical contribution by integrating older adults’ perspectives into the conceptualization of DT-based IC assessment. This user-centered orientation was absent from the 5D DT model, which primarily emphasized system architecture, data flows, and computational mechanisms but offered limited consideration of older adults’ acceptance, perceptions, and needs. However, the “windmill” model presented a notable limitation in that it did not address the data and connectivity dimensions included in the 5D DT model. To enhance its comprehensiveness, future work should incorporate expert consultations to refine these technical dimensions and integrate them with the older adults-centered components.

The IC assessment based on DT technology guided by the “windmill” model showed considerable promise. The IC assessment is grounded in multidimensional data collected from older adults, such as physiological parameters, handgrip strength, and daily activity [43-46]. By integrating general population-level knowledge with real-time individual data from wearable devices, DT technology enabled personalized and continuously refined IC assessments for older adults [47,48]. The entire process of data collection and IC assessment results presentation through DT technology reflects the ongoing transition in modern medicine towards a data-driven paradigm [49]. Beyond assessment, DT technology informed by the “windmill” model also showed strong potential for IC-oriented interventions. The study demonstrated that personalized digital coaching presents a highly promising approach for interventions targeting IC [50]. Virtual entity could also be seen as a form of coaching. The study had shown that a virtual entity equipped with expert-level medical knowledge could serve as a personal physician capable of answering patients’ questions around the clock [51]. In addition, an older adult’s DT could serve as a platform for computer-simulated therapeutic experiments, enabling the design of precisely tailored and personalized treatment plans [51].

However, the application of DT to the IC assessment in older adults still faces several major challenges, such as high implementation costs [52], limited usability for individuals with low digital literacy [53], and ethical concerns regarding data privacy and consent [54]. To overcome these limitations, future DT systems are expected to incorporate generative artificial intelligence–based interfaces to simplify user interaction and adopt low-cost wearable sensors and privacy-preserving data protocols to mitigate financial and ethical constraints [51,55,56]. These technological advances also provide a foundation for practical implementation in real-world settings. Take the Chinese community setting as an example, older adults could wear basic wearable sensors, and the collected data would automatically update their individual DTs. Based on these real-time data streams, the DT model could provide targeted guidance for daily IC management, while community health professionals could remotely assess and manage older adults’ IC through the DT system.

There were several limitations to this study. First, this qualitative study sample involved participants from parks and leisure clubs for older adults but did not include older adults from homes, nursing homes, or hospitals. This limited the generalizability of our findings to the broader population of older adults. Second, all interviewees were from urban areas and had some level of familiarity with DT technology. This might have influenced their perceptions and suggests that the findings might not fully represent older adults living in rural areas or those with limited exposure to DT technology [57]. In the future, a more diverse sample should be considered, including participants from different regions and with varying levels of technological experience, to improve representativeness and capture a wider range of perspectives.

To integrate DT technology for mapping the ICs of older adults and enable comprehensive management of ICs, a preliminary conceptual model, termed the “windmill” model, was proposed. According to the “windmill” model, the ideal digital avatars of older adults had diverse personalized characteristics. To help older adults understand their ICs, multimodal reminders and synchronous changes of digital avatars were essential. Furthermore, the DT system was believed to provide various feedback services to older adults in terms of health, daily life, psychology, and society. Older adults also had some expectations for DT technology, such as intelligence, knowledge integration, and positive feedback. These findings provided guidance for the development of DT systems tailored to effectively support IC mapping in older adults, ultimately contributing to healthy aging.