Andrew M. Kahn

This paper introduces Bayesian VFM (B-VFM) to address ill-posedness in vector flow mapping by jointly inferring velocity fields, segmentation masks, and uncertainties using Bayesian inference.

This paper develops two methods to estimate left atrial blood residence time from 4D CT contrast data, showing that indicator dilution theory provides faster, more accurate results than a physics-informed neural network.

This study simulates blood flow and coagulation in patient-specific left atrium models to assess how hemodynamics affect the efficacy of factor XI/XII inhibitors in preventing thrombosis.

This paper introduces AI-VFM, a physics-informed neural network method for super-resolution left ventricular flow and pressure mapping from echocardiographic data.

This study evaluates a precision health platform integrating whole genome sequencing and noninvasive clinical assessments in 1190 healthy participants, identifying significant genetic and phenotypic findings linked to disease risk.

The paper integrates multimodal health data to identify novel biomarker signatures for health and disease, showing better predictive power for diabetes and hypertension than traditional biomarkers.

This paper screens adults using whole genome sequencing and imaging to detect age-related chronic disease risks, finding unrecognized diseases in 8% of participants and genomic correlations in 25%.