Rapid Innovation of Precision Psychiatry Interventions Using Dynamical Systems Modeling and Ecological Quasi-Experiments: Protocol for a Proof-of-Concept Study on Eating Disorders.

Background: Nearly 83 million Americans experience mental illness, but fewer than 50% accessed past-year mental health services. Disseminable, potent mental health interventions are critically needed. Standalone digital interventions offer promise for improving clinical outcomes, but the efficacy of these interventions to date is modest, likely due in part to insufficient personalization to patient heterogeneity, varying context, and the stochastic nature of mental health symptoms. Objective: This study seeks to improve the efficacy of precision digital health interventions by developing real-world, person-specific models for psychopathology using ecological data and dynamical systems modeling. Dynamical systems modeling of ecological momentary assessment (EMA) and smartphone sensor data will provide insight into each individual's maintenance system for psychopathology in their daily life. Just-in-time adaptive interventions (JITAIs) provide support in the moment at specific decision points, often via smartphone, when the risk for maladaptive behaviors is determined to be elevated. This study will use microrandomized assignment to deliver JITAIs at times of risk for maladaptive behavior based on the individual's dynamical system to evaluate the granular effects and mechanisms of specific therapeutic skills. Proof-of-concept of these methods for improving potency of precision psychiatry digital interventions will be evaluated in eating disorders (EDs), as EDs are characterized by easily measurable maladaptive behaviors, maintained by a complex intersection of biopsychosocial factors, and their frontline treatment, enhanced cognitive behavioral therapy (CBT-E), is primarily behavioral. The aims are to (1) test the hypothesis that person-centered dynamical models will describe risk for dietary restriction, binge eating, and compensatory behaviors with ≥60% average area under the curve; (2) test hypothesized effects and mechanisms of CBT-E skills using JITAIs; and (3) characterize feasibility and acceptability of the standalone JITAI system. Methods: The study will enroll 170 adults with EDs. EMA and smartphone sensor data measuring CBT-E risk factors will be collected for 16 weeks. Participants will receive JITAIs delivered based on their idiographic risk models for 10 weeks. ED symptoms, feasibility, and acceptability data will be collected at the end of the study. The study was registered on ClinicalTrials.gov on May 19, 2026. The first participant was enrolled on July 8, 2026. Results: This project is funded from August 2025 to July 2030. Recruitment will occur from June 29, 2026, to July 2029. Data collection started in July 2026 and will end in November 2029. A total of 4 participants have been enrolled (as of August 14, 2026). Results will demonstrate the efficacy of idiographic dynamical systems modeling for accurately describing real-time behavioral risk and provide insight into the relative effects and mechanisms of key CBT-E skills under conditions of daily life. Conclusions: The study will set the stage for a future randomized controlled trial to evaluate the optimized, standalone JITAI, which will have the potential to substantially improve access to evidence-based mental health care.