This study was conducted from April 2022 to February 2023 at the Fudan University Shanghai Cancer Center and Huadong Hospital in Shanghai, China. The purposive sampling method was adopted to enroll participants. The inclusion criteria were as follows: (1) diagnosed with primary breast cancer, (2) aged 60 years and older, (3) undergoing chemotherapy after breast cancer surgery or within 2 weeks of chemotherapy ending, (4) expected survival time of more than 6 months, and (5) know about their own disease status and can express their thoughts and options. The exclusion criteria were as follows: (1) patients currently receiving radiation therapy and (2) patients with vital organ dysfunction. The sample size was guided by the principle of data saturation. Interviews were discontinued after a minimum of 10 sessions, once no new themes emerged across 3 successive interviews [26].

Eligible participants were identified by nurse leaders in the breast oncology departments based on predefined inclusion and exclusion criteria. After briefly introducing the study aims and obtaining verbal agreement from potential participants to be contacted, the nurse leaders referred the interested individuals to the research team.

Before each interview, participants received a detailed explanation of the study from the research team and signed written informed consent forms. General participant characteristics were then recorded, including age, education, chemotherapy stage, primary carer, and whether living with a primary carer.

Subsequently, one of the authors (YZ) and the corresponding author (FW) conducted the face-to-face interviews. YZ is a nursing undergraduate student who took classes in qualitative study, and FL is a lecturer and experienced qualitative researcher with a research focus on breast cancer and imparted qualitative research methodology in the university. Before the beginning of data collection, YZ was trained by FW, covering the principles of qualitative research, techniques of conducting semi-structured interviews (building rapport, active listening, and probing), and mock interview exercises based on the interview guide. Each interview was audio-recorded and lasted 30‐50 minutes. During data collection, the interviewers summarize and confirm their interpretation of what a participant said during data collection as a member checking to secure the trustworthiness.

Data were collected through semi-structured interviews using open-ended questions. The study aimed to gather qualitative data about participants’ behavior and patterns for symptom management and preference for using mHealth; therefore, the questions were developed carefully around this topic. The original questions were refined after trial interviews with 3 potential patients, and the final interview outlines are presented in Textbox 1. The data from trial interviews were not included in the final data analysis.

This study was reviewed and approved by the Institutional Review Board of Fudan University School of Nursing (IRB no. TYSQ2021-03-05). All participants provided informed consent prior to data collection.

The audio-recorded interviews were transcribed within 24 hours after interview completion. All transcripts were managed and coded using NVivo 12. Data were analyzed using qualitative content analysis, and transcripts were coded inductively [27]. By using inductive coding, the data drove the analytical process, ensuring a comprehensive exploration of patterns and themes based on the interview transcripts. This approach allows the researcher to derive insights and meaning directly from the data rather than imposing preconceived coding categories. YZ and DL analyzed the data independently, and the main steps are as follows: (1) immersing in the data by line-by-line reading of the transcript; (2) selecting the unit of analysis; (3) making sense of the data as a whole; (4) open coding; and (5) group and categorization. After open coding, the preliminary codebook was developed and was iteratively refined through constant comparison and discussion between 2 researchers. Codes were then clustered into subcategories and higher-level categories. Intercoder reliability was addressed through regular meetings to compare coding results and resolve discrepancies via discussion until consensus was reached. Investigator triangulation was implemented by involving 2 researchers in data coding and theme development. Discrepancies were resolved through discussion with the other 2 researchers (CY and FW, with 25 y and 10 y of qualitative research, respectively) to reach consensus. Investigator triangulation was used to ensure a comprehensive and trustworthy exploration of subcategories and categories grounded in the interview transcripts.

After the qualitative data analysis, we identified the categories and subcategories of symptom management behavior and preference for using mHealth to manage symptoms, which could be used as the persona segmentation template. Patients with similar features of categories and subcategories were grouped. Based on these groupings, a persona was created to represent the shared patterns within each subgroup [28]. This ensured that each persona reflected not just individual traits but collective tendencies observed across participants. These personas were sent back to participants for validation to ensure an accurate representation of their characteristics and experiences. Final groupings were refined through team discussions and reviewed by qualitative experts to ensure consistency.

Subsequently, we designed the persona skeleton, a layout describing the key information that the persona includes [28]. After finalizing the details of each persona, we used visual icons to highlight their key features and enhance vividness. To further humanize the personas, we assigned them names and incorporated representative quotes to reflect each target group’s personality and context [25]. The visual elements were hand-drawn by the research team to maintain authenticity and engagement.

A total of 17 participants were approached and agreed to participate in the interviews. Participant characteristics are outlined in Table 1. The participants had a mean age of 69.6 (range 61‐80) years. All the participants were female and had undergone radical mastectomy. The most common primary caregivers were their adult children.

We identified 3 categories, 7 subcategories, and corresponding descriptions (Table 2). The results describe the symptom management behaviors and preference for using mHealth to manage symptoms among Chinese older adult breast cancer patients.

We classified our findings into 5 personas, with differences in social interaction patterns, mHealth literacy, and symptoms. We typologized the personas as follows: Positive Manager, Dependent Parent, Management Isolationist, Image Manager, and Clinician-Dependent (Table 3).

Quotes: I think I can battle with breast cancer, and I will eventually win. Long days are there (Textbox 2).

Quotes: I don’t know what to do and have lost hope for the future. Please tell anything to my children (Textbox 3).

Quotes: I am so severely ill. I don’t want to say anything (Textbox 4).

Quotes: I haven’t retired. I have to go back to work with a normal body shape (Textbox 5).

Quotes:

I only rely on health care providers, and I want to contact doctors when I need (Textbox 6)

This study depicted the differences in symptom management behaviors and preference for using mHealth to manage symptoms among the older adult breast cancer patients in China and, based on the differences, built 5 typical personas: Positive Manager, Dependent Parent, Management Isolationist, Image Manager, and Clinician-Dependent. This study outlines the complete process of persona development, including qualitative data collection, persona segmentation, and creation, building upon previous reviews of persona technologies [28]. The strict adherence to the consensus process [28] and early user involvement [29] suggests that our findings may serve as a valuable reference for future persona development and user-centered design.

The qualitative interviews we conducted focused on our target and special population—older adult breast cancer patients. The combination of cancer and aging renders this population particularly vulnerable, as cancer impacts their physical health while aging contributes to increased comorbidities and a decline in overall resilience [30]. Despite previous studies that have primarily focused on the treatment and survival of this population, few have closely examined their self-management behaviors or actively engaged them in the information age to ensure they fully benefit from modern technologies [31]. Notably, we found only 1 previous paper that recognized the need for health technology in chronic disease management [21], and no existing literature specifically addresses the preferences of older adult breast cancer patients. Therefore, our study identifies 3 key categories—interaction patterns, mHealth literacy, and symptom diversity and severity—that revealed the self-management behavior patterns and preference of mHealth for symptom management of older adult breast cancer patients. This research addresses a critical gap in the literature, highlighting the importance of understanding these factors to enhance mHealth solutions for this vulnerable population.

Following this, we developed 5 personas based on the categories and subcategories identified. This approach adheres to standard persona construction practices by using data-driven classification [28]. Moreover, the personas expand traditional persona content to include social dynamics, mHealth access, mHealth usage, mHealth interest, symptom diversity, symptom severity, specific needs, and design points to enhance richness in understanding older adult breast cancer patients and their context [32]. According to previous studies, the enriched insights gained provide valuable guidance for functional requirements, design, and implementation strategies tailored specifically for this target population [21]. The structure and content of our personas, which incorporate the key features of symptom management, can serve as templates for developing personas for other health consumer target populations [21].

We believe that the personas developed can effectively highlight both common and distinct traits among users. The shared characteristics of the 5 personas should serve as the primary focus for core functionalities, while the unique traits will inform the development of supplementary and optional features. From our research findings, we have initially identified 2 core functionalities: symptom reporting and educational materials.

The symptom self-reporting feature will enable users to log their symptoms daily, facilitating real-time monitoring and proactive management of potential side effects, which is crucial for breast cancer management for older adult patients [33]. For instance, for the Positive Manager, visualizing longitudinal symptom assessments could enhance their understanding of health trends over time. In the case of Dependent Parents, the mHealth product could facilitate a dyad approach where patients and their children both engage with the system, allowing symptom reports to be sent directly to their children via push notifications.

Meanwhile, the educational materials will provide tailored information to enhance users’ understanding of their condition and treatment, empowering them to make informed decisions regarding their health. For the Clinician-Dependent, it is crucial to integrate a voice icon and profile icon, enabling medical advice to be easily accessible and conveyed by a trusted physician. Regarding the educational features of the mHealth product, Positive Manager and Imanage Manager are inclined to learn and manage their health and could benefit from self-directed reading materials and a patient support network, while Clinician Dependent, who prefers individualized attention from health care providers, could be offered one-on-one video appointments. Given that the target population consists of older adult patients who often experience memory and comprehension difficulties, the development team must carefully manage the amount of information presented to avoid cognitive overload. A modular approach can effectively cater to users, providing essential knowledge for those who prefer not to delve deeper while allowing for broader exploration through optional modules. Active information seekers, such as the Positive Manager and Image Manager, can benefit from this modular design, which ensures continuous content updates and relevance.

As for the role of personas in the design process, they serve as educational tools for design and usability teams. They provide insights that convey project progress to sponsors and offer a “word picture” that familiarizes team members with target user groups. Personas structure user typification, enhancing communication within development teams. As noted by Hill and Barteck [34], once established, personas become the primary medium for design discussions, aiding in resolving group conflicts by referencing user perspectives. For instance, team members might ask, “What would the Clinician-Dependent think?” while designing the interface and algorithm. Additionally, based on interaction design principles like the double-diamond design, the iterative process often involves usability testing [35]. Personas can help predict who the primary and secondary users will be. For breast cancer patients, spouses and children are typically considered secondary users, as they can facilitate adoption and usage and are often willing to assist primary users in receiving better care. Therefore, detailed personas can serve as a foundation for selecting and screening participants for interactive prototyping and usability testing, directly influencing the solution’s design and development [21]. For instance, when considering Dependent Parents, it is crucial to involve their children as secondary users of the mHealth tool, as their experiences and insights are equally valuable.

During the implementation stage, personas serve as essential tools to guide the deployment of mHealth solutions. For the Positive Manager, promotional efforts such as posters and advertisements can effectively reach this group. For the Image Manager, word-of-mouth within social circles is a key avenue for adoption. The Clinician Dependent persona can be introduced to mHealth solutions through doctors or nurses during appointments, making it crucial to integrate mHealth into the clinical workflow. For the Dependent Parent, the focus can be on caregivers who accompany them to appointments, using this relationship as a key promotional touchpoint. Additionally, for long-term adherence, especially with Dependent Parents, their children should be considered the primary channel to improve compliance and ensure ongoing engagement with the mHealth tool.

According to previous studies, personas do not need to be recreated from scratch for every application [36]. Our 5 specified personas—The Positive Manager, The Dependent Parent, The Management Isolationist, The Image Manager, and The Clinician Dependent—can be reused by both researchers and practitioners. With China’s rapidly aging population and increasing prevalence of cancer, innovative approaches to patient management are urgently needed [37]. These personas offer valuable insights that can address the current gap in China’s user-centered design practice, particularly in mHealth solutions tailored to older adult breast cancer patients. Their reusability makes them adaptable for various health care applications, enhancing design efficiency and consistency across future projects.

Our study presents some limitations that should be acknowledged. First, we relied on interviews for data collection, without using focus groups. While one-to-one interviews provided in-depth insights into individual experiences, focus groups could have uncovered collective perspectives and group dynamics, particularly valuable when studying older adult populations. Second, the personas we developed are based on qualitative data, which, while rich in detail, are inherently subjective. The personas were constructed by researchers based on interpretations of the qualitative data, which may introduce some bias. Quantitative approaches to persona development, such as surveys or statistical analysis, could offer more objective, testable profiles that may provide additional validity and generalizability to the findings [38].

This study identified symptom management behaviors and preferences for managing symptoms using mHealth among older adult Chinese breast cancer patients. Further, we developed 5 personas that reflect the diverse preferences of this population. These personas play a crucial role across multiple stages of mHealth development: helping to determine core functionalities during the planning phase, guiding design and development by aligning product features with user needs, and supporting implementation by facilitating adoption and engagement strategies. Together, these insights contribute to more effective, user-centered mHealth solutions.