We recruited 2 administrators and 3 caregivers. The inclusion criteria for the participants were (1) more than 2 years of nursing experience and (2) current participation in life story intervention in older adult care. Table 1 displays the characteristics of the participants in acquiring the initial design requirements.

We conducted a follow-up observation, supplemented by 5 semistructured one-on-one interviews during the observation [16,17]. The follow-up observation was to observe the 3 caregivers interacting with older adults in clinical care and the way they organized and looked up their life stories. The 5 semistructured one-on-one interviews were conducted with administrators and caregivers to gain insight into their difficulties in organizing the life stories of older adults and their design needs for a life story visualization system. Based on the data analysis results of the follow-up observation, we mainly discussed the following questions with the participants in the semistructured interview:

The above tasks were audio recorded with participant consent and transcribed into words by team members to familiarize themselves. The team members then annotated and categorized the data based on the observations, initially identifying distinct user requirement themes [18]. Subsequently, data from 3 semistructured interviews were chosen for in-depth analysis to validate whether the identified themes aligned with the findings from the follow-up observation. Necessary adjustments and enrichments were made to the themes during this process. Subsequently, the theme framework was further validated using data from 2 additional semistructured interviews.

We designed a low-fidelity prototype (Figure 3) of the Story Mosaic system with the initial design requirements. The mapping relationship between initial design requirements and prototype functions is shown in Multimedia Appendix 1. The system completes tasks including organizing stories chronologically, categorizing them according to topics, and creating a timeline. It depicts crucial information from these life stories, giving caregivers information they can use to provide individualized care for older adults.

We discussed and iterated sketches of the system within our multidisciplinary team, which included 2 user experience designers, 2 computer science researchers, and 2 caregivers (C1 and C2, who participated in the initial design requirements study). Once we finalized the initial design of our system, we converted the sketches to a digital low-fidelity prototype using MoDao (a prototyping tool that facilitates collaborative web-based interaction among multiple users).

Given the relatively older age of caregivers [19], we adopted an age-friendly interaction design based on HCD during prototyping [20-22].

Regarding font size and colors, our design used large font sizes and warm colors to foster a friendly interface [23].

We considered features to make the system easy to click. As individuals age, they may experience a decline in hand-eye coordination and motor functions, making it challenging for them to interact with a user interface (UI) [24]. Tasks such as clicking on interface targets, moving between interface elements, and responding to on-screen targets become difficult. We addressed this difficulty by ensuring that the interactive UI components were sufficiently large with a minimum diagonal length of 11 mm and adequately distanced from other elements with at least a 2-mm gap. Additionally, scroll bars could pose usability issues. The system was designed to simplify scroll bars and provide users with alternative options such as clicking on scroll bar arrows, using the keyboard’s up and down keys, and having action buttons that return them directly to the top of the page [25].

Lastly, we aimed to have consistent layouts. We established a well-defined grid system during the prototype design phase to maintain consistent spacing and alignment. We adhered to a unified typography and color scheme while consistently implementing a standardized set of UI elements and components across the entirety of the design [26].

We recruited 6 participants for the workshops on the Story Mosaic system prototype, comprising 1 UI designer, 1 aging-product designer, and 4 caregivers. In total, 2 of the 4 caregivers were participants in the phase-1 study, and the other 2 were new participants. Table 2 displays the characteristics of the participants in the prototype refinement workshops.

The primary objective of these design workshops was twofold: to acquire feedback on the prototype and to identify unanticipated requirements. Before commencing the participatory design workshops, we obtained informed consent and demographic information from the participants. We conducted 4 workshops with the same group of participants, each consisting of 6 participants, to acquire feedback on the content, presentation, and interaction of our proposed prototype. We conducted participatory workshops with the same group of participants every 2 weeks at the nursing home, recording the audio of each session. The initial 3 workshops focused on seeking feedback and suggestions on different parts of the prototype [28]. After analyzing the data gathered in the first 3 design workshops, we designed an interactive medium-fidelity prototype, ensuring that all issues raised during the initial 3 workshops were addressed. Finally, we presented the medium-fidelity prototype in the fourth workshop, followed by a final round of feedback.

We conducted a thematic analysis of the prototype refinement workshops. The primary objective of this analytical process was to synthesize the feedback received concerning our prototype and to unveil latent user requisites that may have been overlooked during the initial phase of the study. We transcribed the proceedings of the initial 3 design workshops, leading to the inception of preliminary themes. Subsequently, we used the records from the fourth design workshop for validation and refinement. Finally, we curated illustrative quotations corresponding to each thematic category and succinctly encapsulated our discoveries. We refrained from presenting quantitative details, such as enumerating the frequency of information occurrences. The objective of this phase was to unveil unexpected insights through an inductive approach to data collection and analysis [29].

We recruited 7 participants to participate in usability testing. Table 3 displays the characteristics of the participants in the usability tests.

We conducted observational studies to better understand users’ natural interactions with the product, monitoring users as they engaged with the Story Mosaic system. Following literature guidelines [31], we adopted an iterative approach for our usability tests, dividing them into batches. The initial test involved 3 participants, and we addressed their predominant usability issues. We then tested the system again with 4 participants. The moderator asked the participants to use different functionalities within the system while reflecting on the experience of using life stories. During the usability tests, the moderator noted any difficulties the participants encountered. The participants filled out the System Usability Scale (SUS) at the end of the usability tests. We deployed the Story Mosaic system in the Hefei Tianyu Senior Living Service Nursing Home for practical use verification. The usability task and interview questions are presented in Multimedia Appendix 2.

We systematically analyzed and categorized user interactions with our system during each task. The Results section details the usability test results and provides the associated optimization recommendations.

The prototype of the Story Mosaic system primarily encompasses functionalities within 3 modules: older adults management, life story management, and system settings. The life story management module is the core of the Story Mosaic system, and we describe it in detail in the following section. The life story management module contains 3 important functions: life story input, automated life story organization, and timeline generation. The mapping relationship between initial design requirements and prototype functions is shown in Multimedia Appendix 1. The low-fidelity prototype of the life story management module is shown in Figure 3.

The life story input module empowers users to manually input pivotal components, including time, location, individuals, and event summaries, with the flexibility of these particulars being nonmandatory. Moreover, the platform provides an algorithmic mechanism for automated padding in instances where the above information is absent. Users can enrich life stories with various formats, encompassing textual content, images, and videos. Furthermore, the system supports theme personalization, enabling users to establish, modify, eliminate, and retrieve the topics for older adults’ life stories within the framework.

In the low-fidelity prototype, we established 5 primary themes: “Friends and Family,” “Career,” “Achievement,” “Marriage,” and “Interests.” It is within the prerogative of users to exercise adaptability, enabling the customization and introduction of novel themes, all contingent on the idiosyncratic life experiences of older adults. The automated life story organization module allows users to classify life stories through manual themes specification. In the absence of specified themes, the system defaults to an algorithm for automated organization. Narratives are methodically sequenced based on temporal attributes, from earlier experiences to the most recent events. Additionally, the life story presentation interface incorporates a search box, facilitating the execution of fuzzy searches across these narratives.

The primary objective of the timeline generation module is to construct a chronological framework of older adults’ life journey, furnishing caregivers with a swift avenue to comprehend the older adult. The core functionality of this module revolves around streamlining details within the older adult’s life narratives, highlighting pivotal events within each life story. Subsequently, the crucial events are organized based on chronological order, generating a cohesive timeline. Each node on the timeline is represented as “Time: Story Summary.” Figure 4 shows an example of the timeline generation process [32].

The participants had valuable insights and feedback on the prototype, which helped us to identify several areas for functional design and improvement during the prototype refinement workshops. We also discovered some unanticipated requirements. We analyzed the results of the prototype refinement workshops to establish the functional architecture of the Story Mosaic system and design the medium-fidelity prototype. The functional architecture diagram is shown in Figure 5. The medium-fidelity prototype of the story visualization, which is the most important module of the Story Mosaic system, is shown in Figure 6. Multimedia Appendix 4 provides the feedback analysis of the workshops.

Visualizing older adults’ life stories is a central aspect of the Story Mosaic system. In the prototype refinement workshops, we extensively discussed the visualization methods with the participants. Considering their work requirements and suggestions, we iteratively optimized the forms of visualizing life stories. Table 4 presents the participants’ perspectives on the strengths, weaknesses, and improvement suggestions for each visualization approach in the prototype refinement workshops.

Participants offered insights regarding the system’s usability and visual design. They indicated a desire for clear instructional cues, such as an evident “click here to return” prompt during page navigation. They also noted difficulties in interacting with the system’s button, primarily when solely represented by icons or text. A preference emerged for buttons combining icons and text, accompanied by an adequately sized font. Moreover, the frequency of retelling life stories carries varying implications for an individual. Stories mentioned often indicate events that had a profound impact or left a strong impression on the older adult, thus forming suitable topics to initiate engagement in care. On the other hand, seldom recounted experiences could signify events the individual prefers not to discuss, suggesting possible areas necessitating tactful handling in their care plan. Responding to feedback from nursing home caregivers who participated in the survey, we adopted the progressive color representation method.

The following additional functionalities were requested. (1) The life story management function was deemed valuable, designed to preserve life stories that have yet to be included in the life story structure. Users can add, delete, and review life stories in the story bank. (2) The memoir function was desired to enable users to choose some life stories from the life story structure and create an electronic memoir for older adults or their family members. (3) An interface for older adults’ children was also deemed beneficial. The system management module of older adults’ care homes could incorporate individual accounts for older adults’ children. This could let them see caregiver efforts for their parents and help strengthen parent-child bonds through exploring life stories.

The Story Mosaic system underwent usability testing in the Hefei Tianyu Senior Living Service Nursing Home. In all, 5 caregivers and 2 administrators collaborated on inputting 1123 life stories from 16 older adults into the system. The usability task and interview questions are presented in Multimedia Appendix 2. In the usability testing feedback, we summarized the following issues: interaction, bootstrap operation, input errors, and complex operation. Multimedia Appendix 5 details the problems encountered by participants during usability testing and our proposed fixes. We analyzed the results of usability testing to design the high-fidelity prototype. Figure 7 is the high-fidelity prototype of the life story organization and visualization, and the home page of the Story Mosaic system is shown in Multimedia Appendix 6.

Some participants raised concerns about whether the interactive functions were identifiable. For example, the search button represented by a magnifying glass icon was not recognized, or instead of clicking the “view” button, they clicked directly on the story to view its details. It was also difficult for some to view the target life stories. To address these issues with interactive operation, we made the following optimizations:

By observing the participants’ actions, we identified opportunities for improvement in the guided processes. For example, participants exhibited prolonged hesitation when returning to the previous page from the current interface, and they overlooked the “one-click return to top” button on the page when scrolling to the bottom while viewing life stories. In response to the usability issues encountered in the guided operations, we implemented the following optimizations:

Some participants made input errors when adding their life stories, such as ignoring the theme selection. Therefore, in all input operations of the system, input prompts, input warnings, and input error prompts were added. We marked the operation prompt as “required*” for all required fields, to indicate that they must be filled in.

To prevent batch upload failures, the system offered format guidelines, sample files, and preupload format verification with informative error messages for any related issues. Furthermore, we improved password handling by eliminating the need for both uppercase letters and symbols, and we provided accessible guidance for password recovery and resetting.

In the study follow-up, participants completed the SUS, an industry-standard tool offering a holistic measure of usability [33]. As shown in Table 5, the first set of SUS scores was 70.00, 75.00, and 75.00 (mean 73.33), achieving a B– on the SUS grade and indicating higher usability than 65% of other websites. The second set of SUS scores was 82.50, 77.50, 80.00, and 82.50 (mean 80.63), achieving an A– on the SUS grade and indicating higher usability than 85% of other websites. These responses indicate that participants felt the tool was intuitive and valuable enough to use frequently.