PROMISE-AD: Progression-aware Multi-horizon Survival Estimation for Alzheimer's Disease Progression and Dynamic Tracking

2026-04-30

AI Summary

This study addresses key challenges in personalized Alzheimer’s disease progression prediction—namely irregular follow-ups, censoring, diagnostic information leakage, and risk calibration across multiple time horizons—by proposing an information-leakage-resistant survival analysis framework. Leveraging ADNI/TADPOLE data, longitudinal clinical visits are encoded into temporal tokens comprising normalized measurements, missingness masks, and trend indicators. A temporal Transformer integrates global context, attention-based pooling, and the most recent visit representation, which is then coupled with a discrete-time mixture hazard model. The framework employs a multi-task loss function incorporating survival likelihood, focal risk, ranking constraints, and smoothing regularization, along with an isotonic calibration mechanism to yield interpretable multi-horizon conversion risk estimates. It achieves state-of-the-art performance, attaining the lowest integrated Brier score (0.085) for CN-to-MCI and the highest C-index (0.894) and near-ceiling 5-year AUROC (0.997) for MCI-to-AD prediction.

Abstract

Individualized Alzheimer's disease (AD) progression prediction requires models that use irregular visits, account for censoring, avoid diagnostic leakage, and provide calibrated horizon risks. We propose PROgression-aware MultI-horizon Survival Estimation for Alzheimer's Disease (PROMISE-AD), a leakage-safe survival framework for predicting conversion from cognitively normal (CN) to mild cognitive impairment (MCI) and from MCI to AD dementia using ADNI/TADPOLE tabular histories. PROMISE-AD converts pre-index visits into tokens with standardized measurements, missingness masks, longitudinal changes, time-normalized slopes, visit timing, and non-diagnostic categorical attributes. A temporal Transformer fuses global, attention-pooled, and latest-visit representations to estimate a progression score and latent discrete-time mixture hazards. Training combines survival likelihood, horizon-specific focal risk loss, progression ranking, hazard smoothness, and mixture-balance regularization, followed by validation-set isotonic calibration for 1-, 2-, 3-, and 5-year risks. In held-out testing across three seeds, PROMISE-AD achieved an integrated Brier score (IBS) of 0.085 $\pm$ 0.012, C-index of 0.808 $\pm$ 0.015, and mean time-dependent AUC of 0.840 $\pm$ 0.081 for CN-to-MCI conversion, yielding the lowest IBS among compared methods. For MCI-to-AD conversion, PROMISE-AD achieved the highest C-index (0.894 $\pm$ 0.018) and near-ceiling 5-year discrimination (AUROC 0.997 $\pm$ 0.003; AUPRC 0.999 $\pm$ 0.001), although some baselines had lower IBS. Ablations and interpretability supported longitudinal change features, fused temporal representations, mixture hazards, cognitive and functional measures, APOE4 status, and recent conversion-proximal visits. These findings suggest that progression-aware survival modeling can provide interpretable multi-horizon AD conversion risk estimates.

Problem

• Alzheimer's Disease
• disease progression
• survival estimation
• multi-horizon prediction
• individualized risk

Innovation

• progression-aware survival modeling
• multi-horizon risk estimation
• temporal Transformer
• leakage-safe design
• mixture discrete-time hazards