Integrating Clinical Data and Genetic Variation to Understand Comorbidity Profiles and Generate Biologically Optimized Phenotypes

Abstract

Collectively, this work has centered around combining clinical and genetic information to provide insights about disease from various scopes of view. Specifically, it has addressed multimorbidity, comorbidity, and heritability through the development or employment of novel approaches that attempt to further the promise of precision medicine. We compared electronic health records across two major healthcare systems: Vanderbilt University Medical Center and Massachusetts General Brigham to establish multimorbidity networks that validated consistency across institutes despite regional differences in demographics and medical practice. We used this resource to explore comorbidities of schizophrenia to determine which ones were most likely to be easily modifiable due to a lack of shared genetic etiology. Thus, indicating other drivers such as adverse treatment effects, unhealthy behavior, or environmental factors. Then we looked at how to optimize heritability of various diseases through different phenotyping strategies using natural language processing of clinical notes in electronic health records. These showed promise in highlighting overlooked small effects of already known variants and provided potential for identifying previously unassociated single nucleotide polymorphisms. The current health care paradigm is centered around treating conditions through specialists in more of a piecemeal than holistic way and our work highlights the benefits of taking these complex relationships into account.