Abstract
BACKGROUND: Patient interaction patterns with self-triage modules in mobile health apps during urgent direct-to-consumer telemedicine consultations remain underexplored, despite their critical role in optimizing virtual care pathways. OBJECTIVE: This study aimed to analyze user navigation behaviors within the screen pathways of a symptom-based self-triage mobile app's algorithm during remote urgent care assessments. METHODS: This observational, retrospective, single-center study analyzed data from users who were aged 18 years and older and who voluntarily sought virtual urgent care through the Einstein Conecta (version 2.0; iOS and Android) at a private Brazilian hospital between May 2022 and December 2023. Patients with incomplete connection records were excluded. User interactions were evaluated based on the number of distinct triage flows accessed per session, the number of screens viewed per flow, the frequency of returns to previous screens, and the time spent within the self-triage module. Descriptive statistical methods were applied for analysis. RESULTS: Data from 62,006 unique users with a mean age of 36.51 (SD 10.53) years, of whom 54.65% (33,889/62,006) were female, were analyzed. They initiated 194,976 self-triage flows. We found that 36.89% (22,875/62,006) of users completed 1 flow per session; 22.15% (13,734/62,006) users accessed 2 flows; and 27.93% (17,317/62,006) users accessed ≥4 flows (maximum 63 flows). Users receiving an initial emergency department referral recommendation were more likely to initiate subsequent flows than those recommended for virtual assessment. Returning to a previous screen was infrequent (used by 5277/62,006, 8% of users). The average time spent in the first flow was 70.95 (SD 65.26) seconds, with an average of 9.51 (SD 12.84) seconds per screen. CONCLUSIONS: In this cohort, most users explored alternative pathways beyond the initial self-triage recommendation, particularly when directed to the emergency department, while rarely backtracking within a flow. These findings underscore the need to refine self-triage mechanisms in telemedicine to better align with observed user navigation behaviors and preferences.