A Washington University team reported an experimental blood test that distinguishes Alzheimer’s, Parkinson’s, frontotemporal dementia and dementia with Lewy bodies, and can detect overlapping disease in the same patient.
The test measures 15 proteins and uses an AI model trained on blood samples and records from more than 3,000 patients to estimate which diseases are present and in what proportion.
Researchers validated the model against a separate Arizona patient group by comparing blood-test predictions with brain tissue findings at autopsy, producing 92.3% accuracy for mixed disease detection.
The approach could improve treatment matching because current blood tests largely focus only on Alzheimer’s, though outside experts said larger, more diverse studies are still needed.
The team is discussing clinical trials and an FDA approval path for GPND-AI, while emphasizing that blood biomarkers would complement rather than replace full clinical evaluation.
A new AI test diagnoses four brain diseases. How will this revolutionize dementia treatment and patient care?
If one blood test can predict incurable brain diseases, what are the ethical lines we must now draw?
AI-Powered 15-Protein Blood Test Achieves Over 90% Accuracy in Early Diagnosis of Major Neurodegenerative Diseases
Overview
A groundbreaking AI-powered blood test, developed by Carlos Cruchaga’s team at Washington University in St. Louis, uses artificial intelligence to analyze a panel of 15 specific proteins in blood. By training an AI classifier on data from over 3,200 individuals collected from leading research centers, the test can accurately distinguish between major neurodegenerative diseases such as Alzheimer’s and Parkinson’s. This innovative approach promises to make diagnosis faster, more precise, and accessible, potentially transforming how dementia and related conditions are detected and managed in the future.