Networked trial emulation framework for causal effects of Glucagon-Like Peptide-1 Receptor Agonists on Mental Health Outcomes

About This Grant

Summary This proposal seeks to develop a networked trial emulation framework to enhance real-world evidence (RWE) generation on the causal effects of glucagon-like peptide-1 (GLP-1) receptor agonists (RAs) on mental health outcomes. While GLP-1 RAs have shown benefits for type 2 diabetes and obesity, their potential impact on neuropsychiatric health remains underexplored due to methodological challenges, including measurement errors, immortal time bias, incomplete drug adherence data, and decentralized multi-institutional data sources. This project addresses three critical gaps: (1) the need for systematic-error-corrected target trial emulation framework to improve causal effect estimation for mental health outcomes; (2) the lack of scalable methods for multi-institutional causal inference in neuropsychiatric research; (3) the challenge of accurately phenotyping drug adherence and mental health conditions using structured and unstructured clinical data. The proposed framework integrates systematic-error correction techniques, federated learning, natural language processing (NLP)-enhanced phenotyping, and tensor train decomposition to overcome these limitations. The success of this project will establish a robust, scalable approach for high-quality RWE generation, advancing the understanding of GLP-1 RAs' potential neuropsychiatric effects and informing clinical and regulatory decisions.