AYACS provided feedback on the mid-fidelity prototype during individual interviews. Individual interviews were conducted via Cisco Webex by KMI. KMI is a female researcher with a PhD in Clinical Psychology and extensive training in qualitative data collection methods, including a certificate in user-centered design. At the time of the study, KMI was an Instructor at the Wake Forest University School of Medicine. She did not have a relationship with study participants before study commencement, although some participants had previously engaged in co-design workshops with KMI as part of the larger study. Participants were informed that the primary purpose of the interviews was to inform the development of a digital tool to help AYACS better manage symptoms of depression. Each interview lasted approximately 1 hour. Transcripts of the individual interviews underwent rapid qualitative analysis to determine what changes should be made to the mid-fidelity prototype. The findings informed the design of a working prototype.

Potentially eligible AYACS were identified through electronic medical record review at a large academic medical center in the Southeastern United States with a National Cancer Institute–designated comprehensive cancer center. At the time this study was conducted, this center served over 1300 AYACS annually. Based on the surrounding county demographics, the average minority distribution of AYACS was estimated to be approximately 13% African American, 2% Asian/Pacific Islander, 13% other/non-White, and 8% Hispanic/Latinx.

Potentially eligible participants were identified from the electronic medical record if they met the following criteria: (1) currently between the ages of 15 and 39; (2) diagnosed with cancer between 12 and 39 years of age; and (3) between 1 month and 5 years after completion of primary treatment for their cancer. Study staff contacted potentially eligible patients or their parents/guardians (for adolescents) by sending a message about the study through the electronic patient portal. Follow-up calls or emails were sent within 2 weeks to nonresponders. Across all modes of contact, we attempted to reach potentially eligible participants no more than 3 times before receiving a response. Recruitment materials directed interested individuals or their parents/guardians to a REDCap (Research Electronic Data Capture; Vanderbilt University) [31] screening form to confirm their eligibility for the study. In addition to evaluating the criteria described above, this form was also used to ensure that the patient (1) was fluent in spoken and written English; (2) owned a smartphone with a data plan; (3) had access to a computer with internet access, a webcam, and a microphone that could be used to participate in the interview; and (4) did not have a current diagnosis of a severe or persistent mental illness. Eligible patients and, in the case of adolescents, their parents/guardians were automatically directed to complete digital assent/consent forms. If eligibility could not be determined based on the screening form alone, study staff contacted the participant to make a final determination regarding eligibility for the study and, if eligible, manually sent the participant the digital assent/consent forms.

Upon providing assent/consent, AYACS were asked to provide demographic and clinical information, along with their availability to participate in study procedures. Study staff then contacted the participant to schedule the interview.

All study procedures were reviewed and approved by the Wake Forest University Health Sciences Institutional Review Board (IRB00075887). Participants provided informed consent or assent before engaging in the interview by reviewing and signing a REDCap form. This REDCap form included a brief overview of the study, information about study procedures, details regarding the risk level of the study (minimal risk), potential risks (eg, breach of confidentiality) and benefits (eg, helping others), compensation information, and ways to contact the principal investigator and Institutional Review Board. For adolescents, parents were initially contacted and asked to provide permission for the adolescent to participate. Study staff contacted adolescents directly only after parental permission was provided.

All data were collected remotely using platforms designated as compliant with the Health Insurance Portability and Accountability Act (HIPAA) and stored on secure, password-protected servers accessible only to study staff. Individual interview recordings were transcribed and deidentified for analysis. All information linking participants to their data will be deleted upon publication of the primary study outcomes.

To compensate participants for their time, a US $40 digital gift card was emailed to the participant following the interview.

All interviews were conducted via Cisco Webex. Participants were asked to complete the interview in a quiet, private space. Only KMI and the participant were present during the interview. At the beginning of each interview, KMI provided context regarding the project goals and the work completed thus far (ie, co-design workshops with AYACS and providers). Most of the interview was dedicated to showing participants the mid-fidelity prototype of the tool, explaining how each page of the prototype might work, and requesting targeted feedback (see Multimedia Appendix 1 for example interview questions and associated slides; questions were not pilot-tested before study commencement). At the end of the interview, participants were also asked to provide any additional feedback that had not been shared during the interview.

Individual interviews were recorded and auto-transcribed using Cisco Webex software. Given these recordings, field notes were not made by KMI, and transcripts were not returned to participants for comment or correction. Transcripts were cleaned by study staff. The original audio/video recordings were consulted as needed for clarification. Our sample size of 14 is within the recommended number of interviews needed to reach saturation in qualitative research [32] and in other work focused on developing digital interventions for young cancer survivors [33].

Guided by the Rigorous and Accelerated Data Reduction technique for rapid qualitative analysis [34], we derived themes from the data. Interview transcripts were organized by interviewee, aspect of the digital tool in question, and specific interview questions in an Excel spreadsheet (Microsoft Corporation) to serve as the phase 1 data table. Next, 2 coders (KMI and GW) collectively brainstormed an overarching question to determine which responses would be included in the reduced phase 2 data table: What changes do we need to make to this prototype to more fully integrate AYA needs and preferences for a digital tool to help them manage depressive symptoms? A response was included in the coding process only if it answered this question directly. The coders independently determined whether each response met this criterion. They then compared and discussed their decisions to form a final list of responses to be included in the coding process. During this process, the coders also identified data they believed would be best captured quantitatively. This approach was most often warranted when the interviewer presented the interviewee with options from which to choose, allowing the interviewee to express a preference. For example, participants were shown several options for what the mood rating scale might look like and were asked how often they would be willing to track their mood. Choosing to capture data quantitatively did not preclude including associated context in the qualitative coding process. Quantitative data were summarized using descriptive statistics.

Next, the coders individually developed preliminary codes for the first 3 interviewees’ responses. They then compared and discussed these code ideas to determine the final codes for each response. Throughout this process, the coders attempted to use as much of the interviewees’ original language as possible to carry forward the interviewees’ sentiment. The coders then repeated this process with the next 3 interviewees’ responses, creating and adding new codes as needed until all interviewees’ responses were coded.

Following the application of codes to all interviewee responses, the coders reviewed the codes and began organizing them into themes and subthemes. Themes and subthemes were derived both empirically, based on the interview transcripts, and a priori using existing coding schemes from related work [35,36]. Coders applied themes and subthemes independently, then compared and discussed them to determine the final themes for each code, returning to the response data as needed. A final version of the data table was created to assist with the identification of major codes and illustrative quotes that helped define, refine, and illustrate the identified themes. Participants were not offered the opportunity to provide feedback on the findings.

Based on feedback and recommendations from the interviews, revisions to the prototype were identified by the study team. BrightOutcome provided reprogramming of an existing digital tool (EMPOWER) [37-39] to meet the technical, functional, and aesthetic preferences of end users based on our co-design process.

Specifically, for each property of a subtheme identified through qualitative analysis, a determination was made as to (1) whether a change was needed (eg, no feature addressing a property already existed within the prototype, the existing tool being adapted, or both; the change aligned with best practices for design); (2) what change should be made (eg, the study team’s recommendation for how to best address a property through a specific adaptation); and (3) whether the change was feasible within the constraints of the budget and timeline of the current project. At times, feasibility also informed decision-making (eg, if an existing feature of EMPOWER addressed a property of the subtheme, additional features were not added). Changes that were deemed ideal by the study team but were not feasible were retained as potential future changes.

In conjunction with developing the digital platform through which content would be delivered, we adapted 4 EBIs for depression (behavioral activation, cognitive restructuring, mindfulness training, and positive psychology interventions) for digital delivery to AYACS in accordance with the first 3 steps of the planned adaptation approach [40]: (1) examine the theory of change; (2) identify population differences; and (3) adapt program content in a manner that retains the core functions of the intervention (eg, for behavioral activation, increasing engagement in enjoyable activities) while allowing for tailoring (eg, using examples and stories to demonstrate how core functions such as engagement in enjoyable activities might look different for AYACS than for their noncancer peers).

Fourteen AYACS participated in the individual interviews. At the time of the interview, AYACS ranged in age from 15 to 38 years (mean 25.7, SD 9.14 years). AYACS were diagnosed with cancer between the ages of 13 and 37 (mean 23.4 years, SD 8.9 years) and were between 1 and 55 months posttreatment (mean 23.7 months, SD 16.1 months). The majority of AYACS identified as non-Hispanic White, female, and heterosexual (see Table 1 for additional details).

Based on insights obtained during individual interviews, a working prototype was developed by reprogramming an existing digital tool (EMPOWER) [37-39] by BrightOutcome, a health care technology company. While much of the working prototype aligned with the design of the mid-fidelity prototype and EMPOWER, modifications were informed by quantitative findings and themes identified in the individual interviews. To promote user engagement, we decided to provide AYACS with daily SMS text message reminders that use an encouraging tone (sent only if the user does not engage with the tool by early evening) and to provide graphical feedback on fluctuations in their mood. Regarding tailored content presentation, AYACS emphasized that content should be authentic and age-appropriate. As such, we partnered with 2 nonprofit organizations (Elephants and Tea and Cancer Rebellion) to integrate real stories from AYACS into the tool in both blog and video formats. Further, we worked with a professional animator to create age-appropriate animated psychoeducational videos. To enhance the perceived usability and accessibility of the tool, all videos include transcripts or closed captioning. We also modified the names of modules and added descriptions to provide users with a clearer understanding of what each module entails. Finally, we worked with a professional designer to improve the design of the user interface by implementing a consistent and coherent theme and using brighter colors throughout the tool. We also included artwork throughout the tool created by a professional artist who is an AYACS. Screenshots of the working prototype, now called ASCENT (which stands for the AYA Survivors’ Coping and Emotional Needs Toolkit), are presented in Figure 1.

In addition to reprogramming the EMPOWER platform based on this feedback, we also adapted 4 EBIs for digital delivery. Consistent with our theoretically and empirically derived conceptual model [30], EBIs were adapted for delivery via 5 modules. Core elements of 4 EBIs (behavioral activation, cognitive restructuring, mindfulness training, and positive psychology interventions) were identified by reviewing established clinical treatment manuals, existing digital tools, and prior empirical findings. Population differences requiring adaptation were identified as the experience of having cancer, age or maturity level, and delivery via a self-guided digital platform. Intervention content was adapted to address these differences while preserving core elements and ensuring that there was no overlap between modules. The Feeling module includes 6 sequential lessons that provide general psychoeducational content about AYACS and depression, resources for identifying and managing a mental health crisis, and an introduction to how the tool’s discussion boards can be used for peer support. Each adapted EBI was included as a separate module of 6 sequential lessons based on established treatment protocols: Doing (behavioral activation), Thinking (cognitive restructuring), Being (mindfulness training), and Positivity (positive psychology interventions). Within lessons, text-based psychoeducational content was adapted into scripts for brief, educational animated videos. Interactive exercises such as Socratic questioning and goal-setting were modified for digital delivery. The content was tailored to AYACS by incorporating stories from other AYACS, using age-appropriate language and design, and incorporating features to promote engagement with the tool that were informed by user-centered design with AYACS. Table 2 provides a more detailed description of current and potential future changes to the tool based on individual interview feedback and the planned adaptation process.

This study aimed to explore the preferences of AYACS for the design and content of a digital self-management tool for depressive symptoms. The findings indicate that AYACS highly value personalization, flexibility, and peer support in digital interventions. Specifically, participants emphasized the importance of tailored content that resonates with AYACS’ lived experiences of cancer survivorship, opportunities for interaction with peers facing similar challenges, and features that allow flexible engagement with the tool. These preferences underscore the need for a tool that is uniquely tailored to the needs of AYACS.

This study provided key insights into the unique needs and preferences of AYACS for digital mental health interventions. First, the findings will inform the development of a working prototype of the digital tool being developed by our team and have broader implications for researchers seeking to develop interventions tailored to this population.

A second strength of this study is the use of user-centered design methods with AYACS. User-centered design has been increasingly recognized as an important strategy for informing the development of digital mental health interventions in the general population [41]. Further, user-centered design methods have been successfully used with young people with cancer to inform the development of health care transition and symptom self-management interventions [42-44]. However, such methods have not explicitly been used to develop digital mental health interventions for AYACS [37,45,46], although AYACS have provided input that has guided the development of these tools. By relying on established methods from the field of human-centered design to guide the inclusion of AYACS in informing the design of our tool, we have gained valuable insights into AYACS’ needs, motivations, and preferences for digital mental health tools. Ultimately, we believe that integrating these insights into our tool will help ensure that AYACS engage with it.

The findings of this study also underscore several important considerations for the development of digital tools for AYACS. Tailoring emerged as a key theme, with participants expressing a strong preference for content that reflects their individual cancer experiences. This aligns with findings from our prior work [30], as well as those from other groups that have identified the importance of relevant, cancer-specific content and peer stories in engaging cancer survivors in digital interventions [47,48]. AYACS’ strong preference for tailored content underscores the necessity of providing not just general mental health advice but contextually relevant information that resonates with the unique challenges of this population.

Another important finding is the need for flexibility in the tool’s design. AYACS expressed a preference for content that can be accessed at their own pace, with variation in the duration and frequency of interaction. While prior work has indicated that AYACS broadly prefer on-demand interventions [16], our findings suggest a broader desire for flexibility. Specifically, AYACS emphasized that it is important for users to be able to access the content most relevant to them at the moment when they are most ready to engage. One question this raises is how best to help AYACS make decisions about what content to access and when. Future research on digital tools should consider using adaptive trial designs such as the sequential multiple assignment randomized trial [27] to determine which interventions may be most effective for which AYACS and at what time, ultimately informing the development of a just-in-time adaptive intervention [49].

The study also highlighted the significance of peer support. Participants emphasized that having the opportunity to connect with others who have experienced similar challenges would be an important feature of the digital tool. This is consistent with prior research suggesting that AYACS place a high value on support from peers who have had cancer. Prior research indicates that peer interactions can reduce feelings of isolation and provide valuable emotional support for cancer survivors [47]. Incorporating social features, such as forums or shared experiences from other survivors, could enhance AYACS’ sense of community and support.

Finally, participants raised important ethical concerns regarding privacy when using the digital tool. In accordance with best practices for digital mental health tools, the working prototype of ASCENT was developed on a HIPAA-compliant platform with several technical safeguards in place to protect participant data. Further, participants are never required to share their personal information or responses with other users on the platform; rather, this is an optional part of ASCENT. While participants did not explicitly raise other concerns regarding the potential risks of ASCENT, this remains an important consideration in the development of such tools. Prior research suggests that while digital mental health interventions are generally considered relatively safe, the occurrence, frequency, and methods for safety data collection vary widely, making it difficult to validate this claim [50]. A recent qualitative study provided recommendations for improving the safety of interventions such as ASCENT, including acknowledging the lack of personalization, providing genuine crisis support, and anticipating and supporting deterioration [51]. Such recommendations will be critical in informing future evaluations of the effectiveness and safety of ASCENT.

While this study provides several important insights, it is not without limitations. Despite efforts to recruit a diverse group of participants by identifying individuals from varied backgrounds during our review of medical records, our sample does not fully reflect the broader population of AYACS in terms of race, ethnicity, gender, and sexual orientation. Future research aimed at tailoring digital health interventions for this population should seek to recruit diverse samples of AYACS via targeted recruitment strategies to ensure the generalizability of findings across diverse AYACS. However, it is worth noting that the sample was broadly representative of the typical age range and common cancer types found among AYACS. Further, while our sample size of 14 was sufficient for the planned qualitative approach, future studies are needed to test the generalizability of these findings in a larger sample of AYACS. This could occur through a stand-alone study (eg, a questionnaire designed to validate AYACS’ preferences for digital self-management tools) or as an adjunct to future trials of such tools. The qualitative methodology of this study could also have been more rigorous. For example, it is best practice to pilot-test interview guides before beginning recruitment. Further, it would have been ideal to obtain participant feedback on transcripts and identified themes and subthemes before finalizing our insights and design decisions, as well as to conduct follow-up focus groups.

This study highlights the key features that AYACS value in digital self-management tools: personalization, flexibility, and peer support. These preferences underscore the need for digital interventions that are not only tailored to individual cancer experiences but also flexible in design to meet the evolving needs of AYACS over time. The broader implication of these findings is that digital health tools, when designed with these elements in mind, have the potential to improve mental health outcomes and AYACS’ overall quality of life. By offering a more personalized, flexible, and socially connected experience, digital interventions can play a crucial role in supporting this often-overlooked group of cancer survivors. Future research should continue to evaluate the impact of integrating AYACS’ preferences into digital tools on promoting sustained engagement with digital interventions, which may ultimately lead to improved health outcomes. For example, in our pilot feasibility study of procedures for a full factorial trial of ASCENT, we will collect data regarding usability, engagement, actual usage, and safety (ie, occurrence of adverse events, deterioration). Further, participants will be asked to provide feedback on the working prototype via exit interviews [52].