The Digital Therapeutic Alliance With Mental Health Chatbots: Diary Study and Thematic Analysis
Bibliographic record
Abstract
BACKGROUND: Mental health chatbots are increasingly used to address the global mental health treatment gap by offering scalable, accessible, and anonymous support. While prior research suggests that users may develop relationships with these chatbots, the mechanisms and individual differences underlying such relational experiences remain underexplored. As the concept of the digital therapeutic alliance (DTA) gains traction, a deeper understanding of subjective relationship-building processes is essential to inform the design of more effective digital mental health interventions. OBJECTIVE: This study aimed to investigate how people subjectively perceive and develop relationships with mental health chatbots over time. We sought to identify key experiential dimensions and interactional dynamics that facilitate or hinder the formation of such bonds, contributing to the evolving conceptualization of the DTA. METHODS: We conducted a 4-week short-term longitudinal diary study with 26 adult participants who interacted with two widely available mental health chatbots (Woebot and Wysa). Data were collected through weekly surveys, conversation screenshots, and semistructured interviews. A reflexive thematic analysis was used to identify recurring themes and interpret the emotional, communicative, and contextual factors shaping participants' relational experiences with the chatbots. RESULTS: A total of 18 participants reported forming a bond or light bond with at least one chatbot. Interview narratives revealed three relational categories: Bond (clear emotional connection), Light Bond (tentative or partial connection), and No Bond (absence of connection). Both participants with lower and higher psychological well-being (based on the World Health Organization-Five Well-Being Index scores) reported forming such relationships, suggesting that bonding capacity is not strictly dependent on mental health status. Thematic analysis identified six key themes that explain why people did or did not form bonds: the desire to lead or be led in conversation, alignment between preferred style of self-expression and accepted inputs, expectations for caring and nurturing from the chatbot, perceived effectiveness of the chatbot's advice and proposed activities, appreciation for colloquial communication, and valuing a private and nonjudgmental conversation. CONCLUSIONS: Our findings provide empirical insight into how people interpret and engage in relational processes with mental health chatbots, advancing the theoretical foundation of the DTA. Rather than favoring one design style, our analysis highlights the importance of alignment between preferences and the chatbot's interaction style and conversational role. Participants' initial expectations around empathy and trust also shaped how relationships developed. Drawing on these insights, we suggest that chatbots may better support early therapeutic relationships by blending emotional support with relevant guidance, allowing flexible input methods, and maintaining continuity through context-aware responses. These features may enhance their therapeutic value and foster stronger relationships.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.013 | 0.024 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.006 | 0.006 |
| Science and technology studies | 0.005 | 0.003 |
| Scholarly communication | 0.003 | 0.003 |
| Open science | 0.002 | 0.004 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.001 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".