Multi-level Latent Variable Models for Coheritability Analysis in Electronic Health Records

Area of Science:

• Genetics
• Biostatistics
• Computational Biology

Background:

• Electronic health records (EHRs) linked with familial data enable large-scale genetic studies.
• Existing methods for heritability and genetic correlation struggle with complex family structures, diverse phenotypes, and scalability.
• Investigating shared genetic influences is crucial for understanding complex diseases.

Purpose of the Study:

• To develop a robust statistical framework for jointly estimating heritability and genetic correlation in EHR-based family studies.
• To address limitations of existing methods regarding familial correlation, phenotype heterogeneity, and computational efficiency.
• To identify shared genetic etiologies between different types of complex phenotypes.

Main Methods:

• Utilized multi-level latent variable models to decompose phenotypic covariance into genetic and environmental components.
• Incorporated both within- and between-family variations to capture complex familial structures.
• Developed iterative algorithms based on generalized estimating equations (GEE) for robust estimation.

Main Results:

• Simulation studies confirmed the consistency and validity of the proposed estimators across various settings.
• Applied the framework to real-world EHR data from a large urban health system.
• Identified significant genetic correlations between mental health conditions and endocrine/metabolic phenotypes.

Conclusions:

• The developed framework offers a scalable and rigorous approach for coheritability analysis in high-dimensional EHR data.
• The findings support hypotheses of shared genetic etiology between mental health and metabolic conditions.
• This work facilitates the identification of shared genetic influences within complex disease networks.