AD risk score for the early phases of disease based on unsupervised machine learning

Introduction: Identifying cognitively normal individuals at high risk for progression to symptomatic Alzheimer's disease (AD) is critical for early intervention. Methods: An AD risk score was derived using unsupervised machine learning. The score was developed using data from 226 cognitively normal individuals and included cerebrospinal fluid, magnetic resonance imaging, and cognitive measures, and validated in an independent cohort. Results: Higher baseline AD progression risk scores (hazard ratio = 2.70, P < 0.001) were associated with greater risks of progression to clinical symptoms of mild cognitive impairment (MCI). Baseline scores had an area under the curve of 0.83 (95% confidence interval: 0.75 to 0.91) for identifying subjects who progressed to MCI/dementia within 5 years. The validation procedure, using data from the Alzheimer's Disease Neuroimaging Initiative, demonstrated accuracy of prediction across the AD spectrum. Discussion: The derived risk score provides high predictive accuracy for identifying which individuals with normal cognition are likely to show clinical decline due to AD within 5 years.