AI blood test helps identify causes of dementia ^{CEFR B2}

Accurate diagnosis of dementia is challenging because Alzheimer's, Parkinson's and related conditions can appear similar and commonly coexist. Researchers report an AI classifier applied to a panel of blood proteins that separates four common causes of dementia from healthy brain aging and can detect when multiple disease processes are present.

The team selected 15 proteins that reflect brain pathology, including validated markers of Alzheimer's disease and proteins linked to synapse and nerve damage and to inflammation. The classifier was trained and tested on blood protein data from more than 3,200 individuals collected by the Charles F. and Joanne Knight Alzheimer Disease Research Center and the WashU Medicine movement disorders section. Model performance was then verified on a separate group of 225 individuals who had cognitive evaluation in life and brain examination at autopsy. The classifier's outputs aligned closely with pathological burden in brain tissue and with clinical presentation, yielding an overall diagnostic accuracy of 92.3% for cases with a single neurodegenerative diagnosis.

The model provided useful insight in uncertain or evolving cases: for people with mild cognitive impairment or ambiguous clinical diagnoses, predictions of Alzheimer's matched the amyloid plaque burden found at autopsy, and it identified Alzheimer-like changes in some people diagnosed with Parkinson's who later developed dementia. The test is not yet ready for clinical use; further validation in larger, more diverse populations and prospective studies is necessary to confirm generalizability and to assess prediction of progression and guidance for treatment.

Potential applications include selecting patients for clinical trials, enabling large population studies, and helping clinicians decide follow-up, specialist referral and treatment strategies. The work was supported by major funders and reported by Washington University in St. Louis.