AI-Based Blood Test Diagnose Multiple Brain Disorders from Blood Sample

Image: Workflow of ProtAIDe-Dx on GNPC (An, L., Pichet Binette, A., Hristovska, I. et al. Nature Medicine (2026). https://doi.org/10.1038/s41591-026-04303-y)

Diagnosing the cause of age-related cognitive symptoms remains challenging because clinical presentations of neurodegenerative diseases often overlap, and multiple pathologies can co-occur early in the course of decline. As a result, distinguishing Alzheimer’s disease from Lewy body disease or other conditions typically requires multimodal evaluation over time.

A blood-based approach could help streamline this process if it can reliably differentiate underlying etiologies and track disease progression. Researchers now report an artificial intelligence method that can detect multiple neurodegenerative diseases from a single blood sample.

Investigators at Lund University, working together with the Swedish BioFINDER study and the Global Neurodegenerative Proteomics Consortium (GNPC), developed ProtAIDe-Dx, an artificial intelligence (AI) model that analyzes protein measurements from blood. Using advanced statistical learning and a joint-learning framework, the model identifies a set of proteins that forms a general signature of neurodegeneration. The learned signature is then applied to classify specific disorders.

The model was built from protein data on more than 17,000 patients and control participants aggregated across multiple GNPC datasets, described as the world’s largest proteomics resource for neurodegenerative diseases. Findings were validated across several independent datasets. The study was published in Nature Medicine on March 31, 2026.

According to the research team, the AI model outperformed earlier approaches and diagnosed five dementia-related conditions from a single blood sample: Alzheimer’s disease, Parkinson’s disease, amyotrophic lateral sclerosis (ALS), frontotemporal dementia, and prior stroke. The investigators also reported that the blood-based protein profile predicted cognitive decline better than the clinical diagnosis and suggested biologically distinct subtypes among individuals sharing the same clinical label. Many people clinically diagnosed with Alzheimer’s disease showed a proteomic pattern more consistent with other brain disorders.

The researchers noted that the implicated proteins may guide follow-up studies into disease-driving processes beyond diagnostic use. Next steps include expanding the proteomic marker set with mass spectrometry to resolve patterns unique to each disease.

“We hope to inch closer toward a blood test that can make a reliable diagnosis across disorders without aid from other clinical instruments,” said Jacob Vogel, assistant professor at Lund University.

Related Links
Lund University