NYU Langone researchers develop AI to detect heart transplant rejection

Researchers at NYU Langone Health have developed an artificial intelligence model that could reduce the need for invasive biopsies in heart transplant patients. Currently, the gold standard for detecting organ rejection is a biopsy, where a small sample of heart muscle is surgically removed and examined for inflammation.

The new AI approach combines electrocardiogram (ECG) readings with common blood tests to identify patients at risk of rejection. In a study involving thousands of ECG readings from over 2,300 adult transplant recipients, the combined model demonstrated high accuracy. Specifically, in a test group, the combined model correctly identified 94 percent of patients who were not experiencing rejection.

This method aims to address the issue of false positives often produced by blood tests alone, which can lead to unnecessary surgical procedures. While the researchers note that biopsies cannot be entirely eliminated, the AI model could help doctors identify which patients require further investigation earlier and avoid unnecessary invasive procedures for those not experiencing rejection.