Developing and Validating a Computational Phenotype for Perimenopause in Large-Scale Electronic Health Records - PROJECT SUMMARY ABSTRACT Perimenopause, the multi-year transition preceding menopause, affects all midlife women yet remains poorly defined in electronic health record (EHR) data. Fluctuations in reproductive hormones during this transition influence sleep, cognition, cardiovascular function, bone metabolism, and mood, with consequences for long- term chronic disease risk. Current research relies largely on age proxies, self-reports, or cross-sectional hormone measures, which are inconsistently captured and insufficient for population-level analyses. This measurement gap prevents the timely identification of modifiable risk factors and delays preventive interventions. To address this limitation, the proposed project will develop, test, and validate a computational phenotype for perimenopause using EHR data. The study has three aims, 1) to generate an expert consensus framework for defining perimenopause in EHRs through a structured Delphi process with clinicians, researchers, and individuals with lived experience, 2) to implement the framework into a computational pipeline, integrating structured elements (diagnoses, labs, medications, procedures) and unstructured clinical notes, mapped to the Observational Medical Outcomes Partnership (OMOP) Common Data Model, and tested in the All of Us Research Program, 3) to validate the pipeline in the University of Rochester Medical Center (URMC) EHR system and assess associations with health outcomes relevant to midlife women. Across aims, the project emphasizes reproducibility, transparency, and portability by sharing algorithms, code, and documentation openly through GitHub, PheKB, and Observational Health Data Sciences and Informatics resources. The innovation of this study lies in establishing a scalable, reproducible, and temporally precise definition of perimenopause in real-world data. This computational phenotype will enable longitudinal analyses linking perimenopause to subsequent health outcomes, strengthen the evidence base for clinical guidelines, and inform clinical decision-support tools for managing perimenopausal symptoms and risks. By filling a critical gap in reproductive health measurement, the proposed work will improve how perimenopause is identified in clinical systems and advance prevention and intervention strategies to improve midlife women’s health.
