Particle.news

Download on the App Store

AI-Analyzed Blood Signature Flags ALS Years Before Symptoms, Study Finds

Researchers report a 33-protein plasma pattern that separated patients from controls with 98.3% accuracy in Nature Medicine.

Overview

  • An international team led by Adriano Chiò and Andrea Calvo in Turin, working with the U.S. NIH, profiled blood from 183 ALS patients and 309 controls.
  • The Olink Explore 3072 platform quantified more than 3,000 plasma proteins, revealing 33 with significantly altered levels in ALS.
  • A machine-learning model built on this signature distinguished cases from controls with 98.3% accuracy and was replicated in an independent cohort.
  • Altered proteomic signals were detectable in samples collected years before clinical onset, indicating a prolonged preclinical phase.
  • The findings remain at the research stage, with authors calling for broader validation, standardized assays, and prospective studies before clinical use.