Robust Diagnosis from EHRs Integrating Physics-based Missing Data Multiple Imputation and Fast Privacy-Preserving Inference for Hemodynamic Models.
Missing data often limits the ability to extract useful information from electronic health records (EHRs). The goal of this project is to leverage the fact that missing information often satisfies mathematical or physical principles to develop innovative model-based imputation approaches, combining models and efficient privacy-preserving learning techniques for large EHR datasets. Computationally efficient algorithms are developed to train numerical models while preserving patient privacy, and the feasibility and practical usefulness of these approaches is demonstrated at a scale that has not yet been addressed in the literature. Additionally, the approaches for predictive numerical models developed for this project will be applicable broadly in various fields.