JMIR Publications

ABSTRACT

Background:

AI-powered digital mental health interventions (DMHIs) are increasingly deployed to support employee mental health, yet evaluations have focused predominantly on pre-post symptom outcomes. The emotional states that unfold within real-world AI-assisted sessions remain largely uncharacterized, limiting understanding of how and when these tools may support users in distress.

Objective:

This study aimed to: 1) characterize AI-derived emotional states and patterns that emerge during AI-supported mental health sessions, 2) characterize and compare reduced-distress and persistent-distress sessions among sessions beginning in AI-derived distress states, and 3) identify conversational features associated with reduced-distress trajectories.

Methods:

This retrospective study analyzed message-level AI-derived emotional state data from 28,648 sessions contributed by 4,463 unique users of Yuna, an AI-powered DMHI. Yuna's emotion detection system, grounded in Semantic Space Theory, generated continuous scores across 32 fine-grained emotional states at the message level. Fine-grained states were aggregated into four categories: distress-associated, cognitive-processing, regulated-positive, and social-interpersonal. Session-level trajectory types were operationalized from start-to-end emotional category transitions. Group comparisons and logistic regression were used to identify characteristics and conversational predictors of reduced-distress trajectories among sessions beginning in distress-associated states.

Results:

Five trajectory types were identified: stable (26.6%; n = 7,679/28,648), reduced-distress (24.4%; n = 7,042/28,648), transitional-processing (21.5%; n = 6,195/28,648), persistent-distress (15.8%; n = 4,568/28,648), and escalating-distress (11.7%; n = 3,386/28,648). Cognitive-processing states were the most frequently observed at the message level (35.8%; n = 130,611/365,088), followed by distress-associated states (32.2%; n = 117,485/365,088). Among sessions beginning in distress-associated states (n = 1,597/28,648), 73.3% (n = 1,170/1,597) demonstrated reduced-distress trajectories by session end. Reduced-distress sessions were characterized by higher proportions of cognitive-processing (d = 0.53) and regulated-positive states (d = 0.67) relative to persistent-distress sessions. In logistic regression, a greater number of unique AI-derived emotional states (B = 0.143, P<.001) and longer session duration (B = 0.008, P=.033) were positively associated with reduced-distress trajectories, while more AI-derived emotional switches were negatively associated (B = -0.032, P=.002).

Conclusions:

AI-derived emotional trajectory data may offer a scalable, low-burden signal for understanding engagement quality in real-world mental health sessions, laying an empirical foundation for process-level evaluation of AI-powered DMHIs. Future research incorporating self-report validation, longitudinal designs, and experimental methods is needed to determine whether these trajectory patterns carry a meaningful signal for user outcomes.