A mixed methods study evaluated an LLM-powered AI health coach for systemic sclerosis (SSc) self-management in 20 adults over 4 weeks, comparing usage patterns and outcomes between high- (n=8) and low-engagement (n=12) participants.
High-engagement participants had notably more goal-setting (23.6 vs 5.1), information-seeking (10.8 vs 3.9), and companionship (7.9 vs 1.0) interactions than low-engagement participants; 100% vs 33% used the coach for companionship. High-engagement participants showed a greater improvement in fatigue (mean change −5.04, 95% CI −9.34 to −0.73), though no statistically significant between-group differences were found.
Very small sample (n=20) limits generalizability; no control group; exploratory fatigue findings are underpowered and should not be interpreted as causal.
Quantitative usage metrics alone (message count, session duration) may underestimate meaningful engagement with AI health tools — clinicians should consider qualitative interaction patterns when assessing patient engagement. Companionship-oriented use may be a key driver of sustained engagement in patients with SSc.
Explore related topics
