Dual-model deep learning for Alzheimer's prognostication

Clinical Snapshot

Key Findings

• P Value: Not reported in abstract

• Effect Size: Seven-fold differences in conversion rates between risk stratification groups

• Primary Outcome: Time-to-conversion from mild cognitive impairment to dementia

• Nnt Or Sensitivity: Superior survival discrimination maintained across all held-out clinical sites

• Confidence Interval: Near-nominal coverage achieved for uncertainty bounds (specific values not provided)

Clinical Application

Requires cerebrospinal fluid biomarker assessment, which may limit routine implementation but is increasingly available in specialist centers Relevant to Australian dementia services and memory clinics. CSF biomarker testing is available through specialized centers. Could inform PBS criteria for disease-modifying therapies when they become available. Aligns with RACGP guidelines emphasizing early detection and intervention planning. Patients with mild cognitive impairment presenting for initial assessment where treatment timing decisions are required

Abstract

Disease-modifying therapies for Alzheimer's disease demand precise timing decisions, yet current predictive models require longitudinal clinical observations and provide no uncertainty quantification-rendering them impractical at the critical first-visit encounter when treatment decisions must be made. We developed PROGRESS (PRognostic Generalization from REsting Static Signatures), a dual-model deep learning framework that transforms a single baseline cerebrospinal fluid (CSF) biomarker assessment into actionable prognostic estimates without requiring prior clinical history. The framework addresses two complementary clinical questions: a probabilistic trajectory network predicts individualized cognitive decline parameters with calibrated uncertainty bounds that achieve near-nominal coverage, enabling honest prognostic communication rather than false precision; and a deep survival model estimates time-to-conversion from mild cognitive impairment to dementia. Using data from over 3000 participants across 43 Alzheimer's Disease Research Centers in the National Alzheimer's Coordinating Center database, PROGRESS substantially outperforms existing approaches including Cox proportional hazards, Random Survival Forests, and gradient boosting methods for survival prediction. Risk stratification identifies patient groups with seven-fold differences in conversion rates, enabling clinically meaningful treatment prioritization. Leave-one-center-out validation demonstrates robust generalizability, with survival discrimination remaining strong across all held-out clinical sites despite heterogeneous measurement conditions spanning four decades of assay technologies. By combining superior survival prediction with trustworthy trajectory uncertainty quantification, PROGRESS bridges the gap between biomarker measurement and personalized clinical decision-making-providing the prognostic timeline that current staging approaches cannot offer.