We’re delighted that Vironix Health’s paper titled “ScoEHR: Generating Synthetic Electronic Health Records using Continuous-time Diffusion Models” was a featured presentation at the Machine Learning for Healthcare Conference last week in New York City.
Ahmed Ammar Naseer conducted this work for his MSc dissertation at the University of Oxford in collaboration with Vironix Health Inc., under the joint supervision of University of Oxford and Vironix Health.
His paper introduces a novel framework called ScoEHR, which uses continuous-time diffusion models for generating synthetic electronic health records (EHRs). ScoEHR is shown to outperform existing methods in synthetic EHR generation on two publicly available datasets, MIMIC-III and the Yale New Haven Health System Emergency Department dataset, across four standard metrics of data utility. Additionally, in a blind clinician evaluation, the synthetic data generated using ScoEHR was indistinguishable from genuine data, highlighting its realism.
For more information about Ammar’s paper and to obtain a copy, please contact Vironix Health.