“Disease diagnostics using machine learning of B cell and T cell receptor sequences”, published in Science on February 21, 2025, is a world-first demonstration of the scientific underpinnings of the Human Immunome Project. The study, led by Maxim Zaslavsky, Erin Craig, Anshul Kundaje, Scott Boyd and colleagues at Stanford and other institutions, introduces a framework called Mal‑ID (MAchine Learning for Immunological Diagnosis). It analyzes B‑cell and T‑cell receptor repertoire sequencing data from 593 people and accurately predicts whether individuals have COVID‑19, HIV, lupus, type 1 diabetes, recent flu vaccination, or are healthy.
This study demonstrates that machine learning can successfully decode B and T cell receptor sequences to diagnose multiple diseases simultaneously from blood samples, providing proof-of-concept validation for the Human Immunome Project’s core mission to build AI models that can interpret the immune system’s comprehensive record of disease exposures to revolutionize medical diagnostics and accelerate drug discovery.
Dr. Eric Topol provides excellent insights on the approach and explains the implications of the study in his substack, Ground Truths.
News
Responses