AI blood test separates major dementia disorders in early study

A blood draw and an algorithm could change how dementia is sorted

Doctors have long struggled to tell apart the major brain diseases that drive dementia. Alzheimer’s disease, Parkinson’s disease, frontotemporal dementia, and dementia with Lewy bodies can produce overlapping symptoms, and more than one disease process can unfold in the same brain at the same time. That makes diagnosis difficult, delays treatment decisions, and complicates enrollment in clinical trials that depend on clearer biological definitions.

A team at Washington University School of Medicine in St. Louis now reports progress on a simpler approach: an AI-based classifier built around a blood test. According to the researchers, the tool distinguished among four common neurodegenerative diseases, as well as healthy brain aging, with more than 90% accuracy. Just as important, they say it could also detect overlapping disease processes, a feature that addresses one of the biggest gaps in current practice.

Why mixed pathology matters

In many patients, dementia does not follow a single, neat label. Someone may be diagnosed clinically with Alzheimer’s disease, for example, while also carrying Parkinson’s-related or Lewy body pathology. Existing diagnostic tools often force a simpler answer than biology provides. That can matter because treatment strategies, prognosis, and research eligibility all depend on understanding what is actually happening in the brain.

Senior author Carlos Cruchaga said the goal was not simply to produce another yes-or-no test for one illness. Instead, the effort aimed to capture the broader pattern of disease activity in an individual patient. In practice, that would move diagnosis closer to the mixed and layered reality of neurodegeneration rather than the one-disease-at-a-time model that still shapes much of clinical care.

What the researchers built

The new system combines blood-based measurements with artificial intelligence to classify disease state. The supplied report says the model can distinguish Alzheimer’s disease, Parkinson’s disease, frontotemporal dementia, dementia with Lewy bodies, and typical aging. It also identifies cases where more than one disease process is present simultaneously.

That is the central advance. Blood tests for neurodegeneration are attractive because they are cheaper and easier to scale than imaging or more invasive sampling methods. Adding AI allows the researchers to look for patterns across multiple signals rather than relying on a single marker. The result, if it holds up in broader testing, could offer a more practical screening or triage tool for memory clinics and neurology practices.

What could change if the findings hold up

A more precise diagnosis earlier in the disease course could reshape several parts of care. Physicians could have a stronger basis for choosing follow-up tests, counseling families, and matching patients to therapies or trials designed for specific mechanisms. Researchers could also use the tool to better separate patient groups in studies, reducing some of the noise that comes from lumping biologically different cases together.

The study also points to a larger shift in medicine: the movement away from symptom-based categories and toward biomarker-guided classification. Dementia care has lagged behind some other fields on that front because the brain is harder to sample directly. A blood-based method that captures several disease pathways at once would be a meaningful step toward more routine precision neurology.

Why caution still matters

The results are promising, but they do not mean clinicians are about to replace standard workups with a single blood draw. The report describes this as a newly developed tool, not an already adopted clinical standard. Accuracy in a study setting can look different when testing expands to broader and messier real-world populations. Validation, replication, and careful clinical integration still matter.

Even so, the appeal is obvious. Dementia diagnosis today is often slow, uncertain, and fragmented. A relatively inexpensive blood test that can sort multiple disorders and recognize overlap would address a practical bottleneck in both care and research. For patients and families navigating symptoms that are frightening and hard to decode, that kind of clarity could prove valuable well before any single test becomes definitive on its own.

This article is based on reporting by Medical Xpress. Read the original article.