Generative AI for Identifying Mechanisms and Intervention Points in the Intergenerational Transmission of Anxiety

Understanding how anxiety is transmitted from parent to child remains one of the most important yet elusive challenges in clinical science. Parental anxiety profoundly shapes children’s emotional development, yet we lack tools to capture these processes as they unfold in real-world interactions. This is a critical knowledge gap and persistent barrier to developing effective interventions. RAISE-AI (Resilience Across Intergenerational Stresses and Emotional Health through AI-Enabled Discovery) applies generative AI to detect and explain how stress, worry, and fear pass between parents and children. We will develop models that identify when and how anxiety signals are transmitted through multimodal cues (e.g., language, prosody, facial affect, posture) during naturalistic interactions and evaluate these models for accuracy, interpretability, and fairness across racially, ethnically, and socioeconomically diverse families. Finally, we will generate outputs (e.g., summaries, real-time feedback) to inform pediatric anxiety prevention and treatment plans. If funded, RAISE-AI will reveal the mechanisms through which parental anxiety shapes children’s emotional responses and establish one of the first pediatric benchmarks for generative-AI performance in mental health. By translating subtle emotional signals into interpretable feedback, it will help providers support families and advance ethical, culturally responsive pediatric mental health interventions. Key Words: Anxiety; Parent-Child Interaction; Developmental Psychopathology