This page was translated automatically from Turkish. Read the original

This page is an AI-assisted Turkish summary of the study presented at the 26th World Congress of Psychiatry. The study belongs to the presenting authors; please refer to the original source when citing it.

EV0458

Evaluation of an AI-powered real-time stress and mental health management platform for resident physicians under high stress at Los Angeles General Medical Center

EVALUATION OF AN ARTIFICIAL INTELLIGENCE ENABLED REAL-TIME STRESS AND MENTAL HEALTH MANAGEMENT PLATFORM IN A HIGH-STRESS PHYSICIAN RESIDENT POPULATION AT LOS ANGELES GENERAL MEDICAL CENTER

The study evaluating an AI-powered real-time digital health platform that integrates wearable sensor, clinical scale, and chatbot data from resident physicians is ongoing, and no outcome data were presented in the poster.

The study aims to evaluate a context-aware, AI-powered real-time digital health platform incorporating multimodal intelligence for stress and burnout management in a high-stress population of resident physicians; its rationale cites 2023 data from the Association of American Medical Colleges (AAMC) indicating that 48% of physicians experienced at least one symptom of burnout. The clinical outcome and AI system validation study is being conducted over 12 weeks with resident physicians performing active clinical duties at Los Angeles General Medical Center (n = 120), has undergone the University of Southern California institutional review board (IRB) process, and plans to obtain informed consent from participants for access to wearable sensor, smartphone app, and web app data. The assessment uses clinically validated digital scales: psychological capital (PCQ-10), depression (PHQ-9), anxiety (GAD-7), and stress (PSS-10); wearable device measurements, including heart rate, sleep duration, physical activity, and body temperature, are also continuously processed using a deep learning architecture based on a Long Short-Term Memory (LSTM) network and converted into a well-being score. The third data stream involves analyzing conversations with an embedded chatbot for patterns of distress using Large Language Models (LLMs); scores from all data streams are combined using a weighted deep neural network algorithm to generate a single well-being score, with a generative AI-based feedback loop and rewards for positive behavior grounded in behavioral economics. The only quantitative validation result reported in the poster is a proof-of-concept analysis for detecting the onset of anxiety and depression in a real-world data pipeline comprising 27,651 participants; the neural network model achieved approximately 90% accuracy, and this result was presented at the 2025 American Psychiatric Association Annual Meeting. Outcome data from the study itself have not yet been presented; the poster states that the technology is expected to prevent occupational stress and burnout among physicians and improve mental and physical well-being and job satisfaction.

All WCP 2026 summaries