Polypharmacy Community Detection

Polypharmacy Community Detection seeks to improve how medication data are used in studies of aging by moving beyond simple counts of drugs. The project applies graph community detection—drawn from social network analysis—to identify groups of commonly co-prescribed medications across electronic health records and longitudinal cohorts. These medication communities are intended to serve as reduced, interpretable variables for predictive modeling of outcomes, including cognition. The effort includes development of visualization methods to clarify large, heterogeneous pharmaceutical datasets and packaging of code for broader research use.The team is led by Raha Dastgheyb, an Instructor in Neurology at Johns Hopkins School of Medicine with a background in biomedical engineering and computer science. Her work spans analysis of longitudinal cohorts to identify predictors of cognitive impairment in neurodegenerative disease, development of methods for neuronal signal analysis from multi-electrode arrays, and creation of visualization tools and graphical user interfaces to enhance interpretability and reproducibility for researchers. A postdoctoral fellow with expertise in computational neuroscience contributes additional analytic capacity. Together, the team focuses on data science techniques that reveal clinically meaningful medication patterns to inform aging and brain health research.