AURORA
Adaptive Understanding through Real-time Observation, Reporting, and Assistance
Eric Storch · Baylor College of Medicine · United States
What they're working on
Depression acuity can rapidly shift, yet weekly assessments detect change slowly and often unreliably, delaying treatment personalization and compounding disability. AURORA (Adaptive Understanding through Real-time Observation, Reporting, and Assistance) pairs passive sensing with voice-native, conversational GenAI check-ins to capture rich context plus objective multimodal signals. This enables sensitive, safe, and equitable measurement—the foundation for just-in-time adaptive interventions that address stakeholder concerns about privacy, context, and the risk of dehumanizing, metric-driven care. AURORA’s passive triggers launch brief, adaptive audio-visual assessments of depressive symptoms and provide psychoeducation and CBT-inspired engagement prompts/cues, while collecting high-quality audio–visual signals and lived context. The result is an actionable, valid depressive symptom measurement that can individualize treatment. Following development and refinement, we will evaluate AURORA in a pragmatic cluster-randomized trial of 1,000 patients through 100 clinician dyads across 4 continents. With an international team of researchers and lived experience experts, we leverage our strengths in GenAI, digital phenotyping, global clinical studies, ethics, and advanced analytical methods. We aim to: (1) Develop AURORA; (2) Validate AURORA as a psychometrically robust, responsive, and fair measurement system for depression; (3) Demonstrate engagement, acceptability, and clinical utility; and (4) Explore AURORA’s impact on clinical outcomes relative to treatment-as-usual.
AI approach
Generative AI · conversational AI · passive sensing · multimodal measurement
Mental health focus
Depression · measurement · adaptive engagement