WashU Medicine's AI classifier, trained on plasma proteins from over 3,200 individuals and validated against 225 autopsy-confirmed cases, achieved 92.3% accuracy distinguishing Alzheimer's, Parkinson's, frontotemporal dementia, and Lewy body dementia while flagging mixed pathologies. This observational cohort analysis, published in Alzheimer's & Dementia, outperforms single-marker approaches but lacks RCT-level evidence for clinical outcomes or prospective screening. Related work in Nature Medicine (2023) on plasma p-tau217 showed similar early-detection signals in 1,000+ participants yet highlighted racial disparities in biomarker performance, a gap unaddressed here. A 2024 JAMA Neurology review of blood-based diagnostics further notes that while NfL and GFAP panels scale cheaply, real-world implementation hinges on diverse validation cohorts absent from the WashU data. The original coverage underplays how autopsy correlation, though strong, cannot yet predict treatment response years pre-symptom; conflicts of interest around WashU's NeuroGenomics Center also warrant scrutiny. Overall, this signals a diagnostic shift but risks over-optimism without equity and longitudinal data.