Improving the Performance of Models Trained on Small EHR-Derived Samples by Leveraging External Data with Continual Learning Methods.

Jonathan Hui¹, Meng Xia², Jonathan Wilson³, Elliot D Hill³, Abby Scheer³, Lauren Franz³, Matthew M Engelhard³, Benjamin A Goldstein³

¹Jonathan Hui moved to (present address) University of Pittsburgh School of Public Health, Pittsburgh, PA, 15261, USA, after receiving his master's degree in Biostatistics in 2025 at Duke University School of Medicine, Durham, NC, 27705, USA.

²Dr. Meng Xia moved to (present address) Rockefeller University Data Science Platform, Rockefeller University, New York, NY, 10065, USA, after receiving her PhD degree in Electrical and Computer Engineering in 2025 at Duke University, Durham, NC, 27705, USA.

³Duke University School of Medicine, Durham, NC, 27705, USA.

PubMed PMC DOI PDF

Abstract

The performance of an EHR-based deep learning model trained on a small sample can be improved if more data is collected. Instead of collecting more data, the model can be trained on additional data from an analogous external source. However, this risks the model learning patterns in the external data that do not generalize to the target sample. Furthermore, data use agreements often prohibit combi…

Create a free account to keep reading

Free members get 10 full research views every month across publications, clinical trials, FDA clearances, adverse events, and NIH grants. No credit card required.

Create Free Account Sign In

Want unlimited research access? See Pro plans

Data Accuracy Notice: Research intelligence on Health AI Central is aggregated from public sources (PubMed, ClinicalTrials.gov, FDA, NIH, CMS, and others) and refreshed nightly. Classifications and derived metrics are produced by automated methods described in our Methodology. We recommend verifying critical data points against the primary sources before making decisions.