Digital mental health (DMH) tools offer convenient access to resources and strategies for managing psychological wellbeing [9, 86]. However, a common concern is the mismatch between interventions and users' needs at specific times [31], especially with push-based tools like text messages and notifications. While these tools offer potential support at opportune moments, they risk being perceived as insensitive to the user's mental state or availability [31, 87], leading to frustration and disengagement [61, 94]. Incorporating user context information addresses this challenge, enabling DMH systems to deliver timely, appropriate, and relevant interventions [48, 80, 88]. While 'context' is broadly defined [20, 22], HCI researchers often categorize it into contextual factors (e.g., location, time of day, activity level). These factors can be calculated (e.g., time of day) [6, 75], gathered through sensors and digital traces (e.g., location, movement) [48, 80], or actively reported (e.g., mood, energy) [27]. Just-in-time adaptive interventions (JITAIs) use these factors in algorithms and machine learning models to personalize interventions [35, 87]. Despite the promise of JITAIs for sustaining engagement and personalization, they often fail to account for the dynamic changes in people's lives [118, 129], particularly in mental health where factors like social interaction and movement have complex, individualized relationships with mental state [63, 89, 106]. Many DMH tools are designed top-down by experts without nuanced understanding of users' perspectives or preferences. User involvement is often limited to content generation [53], rarely encompassing how context shapes their needs and receptivity to support. Therefore, formative investigations with users are crucial to understand their needs and how they shift over time [22, 53]. This research addresses this gap by investigating users' perspectives to inform the selection of contextual factors in context-aware DMH tools and how dynamic factors impact receptivity to interventions. The study focuses on text messaging systems, a ubiquitous and accessible platform for promoting psychological wellbeing [25], with existing systems adapting based on factors like time of day and activity level [51, 94]. The research questions are: RQ1: Which contextual factors are perceived to influence the user experience of a text messaging service for psychological wellbeing? RQ2: What specific elements of text messaging interventions need to be tailored to reflect the users' dynamic contexts?

This paper first explores the use of text messages for behavior change and mental wellness, then delves into previous research on context-personalized interventions. Text messaging is widely used to promote health behavior change [111] and has shown success in various physical and mental health areas [41, 42, 120, 133]. Examples include reducing alcohol consumption [120], improving medication adherence [34], promoting smoking abstinence [69], weight management [21, 115], and physical activity promotion [56, 85, 116]. In mental health, many text messaging services focus on therapeutic approaches like CBT, DBT, ACT, and motivational interviewing [1, 16, 50, 61, 64, 84, 99, 119]. These services vary in content (motivational quotes, exercise recommendations, peer experiences, reminders) and outcomes. Context-aware computing aims to integrate computation into our environment [22, 82, 109], but 'context' is defined differently. Dey [20] defines it as information characterizing the situation of an entity, while Dourish [22] sees it as a relational property. For technological characterization, researchers often break context down into factors like location, time of day, and activity level [91]. This work focuses on dynamic contextual factors influencing user experience with text messaging for psychological wellbeing, contrasting them with relatively stable contexts like race or profession. Poorly adapted interventions can be irrelevant or inappropriate, leading to content being ignored or negative user experience [87], potentially causing users to quit [30, 66]. Prior research highlights the importance of adapting to users' contexts [62]. Psychological literature links mental health to circadian rhythms [6, 28, 75, 128], time of day [38, 8, 122], physical activity [49, 57, 121], and social interaction [117]. Just-in-time adaptive interventions (JITAI) leverage contextual information to determine intervention timing and type [87]. Examples include adaptive apps based on goals, step count, and environment [51], stress management tools triggered by elevated heart rate [17, 48], and interventions based on self-reported depression and sensor data [94]. However, these often lack user-centered design, selecting contextual variables based on theory instead of user experience [53]. This study addresses this gap by examining qualitative data to identify important contextual variables and how they impact users' experiences and receptivity to interventions.

The investigation involved a formative study and a deployment study. **Formative Study:** This study involved 36 participants (30 recruited via Mental Health America (MHA), 6 from a university) to understand anticipated contextual factors influencing engagement with a text messaging system for psychological wellbeing. Participants exhibiting moderate depression and anxiety (PHQ-9 and GAD-7 scores ≥ 10) were recruited through MHA screening surveys, while university students were recruited through snowball sampling and word-of-mouth. Data was collected through semi-structured interviews and focus group discussions. Interview questions explored contextual factors affecting receptivity to messages, preferred message types, ideal delivery times, and desired message frequency. Thematic analysis was used to analyze transcripts, identifying dominant themes related to daily schedule and affective state. **Ethical Considerations (Formative Study):** Ethical considerations were addressed by informing participants of their right to skip questions or discontinue the conversation. Interviewers received training on the Columbia-Suicide Risk Assessment protocol [96] and safety planning procedures. **Deployment Study:** This study involved 42 participants (some from the formative study) who engaged with two interactive text messaging dialogues: one focused on daily schedule and the other on affective state. Recruitment involved snowball sampling and targeted MHA website ads. Participants received daily messages for 1–2 weeks, engaging with the dialogues focusing on contextual factors on one or two days. The daily schedule dialogue sent two message sequences daily (9:00 AM and 4:30 PM), each including brief activities or reflective questions, followed by a feedback check. The affective state dialogue started with a check-in message assessing mood and energy levels, followed by passive supportive texts (for low mood/energy) or active writing (drafting messages for others) (high mood/energy). Data was analyzed using mixed methods: quantitative response rates and qualitative thematic analysis of interviews to gather feedback on the dialogues and the impact of contextual factors on message receptivity. **Ethical Considerations (Deployment Study):** Participants were informed that the messaging program wasn't a crisis service. Daily review of messages and training on the Columbia-Suicide Risk Assessment protocol were implemented for participant safety.

**Formative Study:** Two dominant contextual factors emerged: daily schedule and affective state. Participants' preferred message times varied, with some preferring mornings for motivational messages, while others preferred afternoons/evenings for support after work. The middle of the workday was generally considered unsuitable, except for low-effort messages. Affective state significantly impacted message preferences, with some participants wanting messages only during low moods, while others valued messages during positive moods. During low moods, simple check-in messages, coping strategies, and peer messages were preferred, avoiding high-effort requests. **Deployment Study:** The deployment study supported the formative study's findings on daily schedule and mood. Participants reiterated preference for morning or afternoon messages based on individual schedules and emotional state. However, it revealed a nuanced perspective on messages during work hours: low-effort reminders or suggestions were appreciated, but high-effort tasks were unsuitable. Participants also suggested the value of strategically timing messages across the day (e.g., background information in the morning, reminders in the afternoon) and avoiding inconvenient times entirely. The affective state dialogue showed generally positive reception, with participants valuing emotion labeling as helpful for emotional awareness and personalized support. However, opinions varied on the number of questions for emotion labeling, with some finding it overwhelming. Participants also suggested a need for more nuanced emotion options and less repetitive system acknowledgement of their responses. Finally, associations between daily schedule and affective state were evident, with mood and energy fluctuating across the day and influencing message receptivity. Participants indicated reduced receptivity when surrounded by others.

The findings address both research questions. RQ1 is answered by identifying daily schedule and affective state as crucial contextual factors influencing text messaging engagement for psychological wellbeing. This supports previous work on time-sensitive interventions and mood-based support. The study extends this literature by revealing diverse perspectives on how these factors shape receptivity and preferences. RQ2 is addressed by identifying message volume, required engagement effort, and time sensitivity as key elements needing adaptation. The need for varied messaging strategies based on mood and energy is confirmed. These findings highlight how users perceive the interplay between their schedules and emotional states. The discussion of design tensions illustrates how users' expectations regarding context-aware DMH tools shift over time. The study contributes to adaptive DMH tool design and context-aware computing by highlighting the importance of gathering user-centered data. It shows how message volume, effort required, and time sensitivity should be tailored to match users' contexts, offering practical recommendations for designing JITAIs. The work emphasizes the significance of formative investigations in understanding the nuances of users' lives and their needs for context-aware support.

This study provides valuable insights into designing context-aware DMH tools using text messaging. The key contributions are the identification of crucial contextual variables (daily schedule and affective state), the identification of specific messaging elements needing adaptation (volume, effort, time sensitivity), and design considerations for incorporating contextual information. Future research could explore different populations, message types, emotion models, and the impact of interventions on symptom reduction.

The study focused on young adults in North America, limiting generalizability. The deployment study used specific message dialogues and message types, which may not fully capture the range of users' preferences. The emotion labeling approach relied on the circumplex model, which has inherent limitations. Finally, the study did not directly assess the impact of the interventions on symptom reduction.