Search papers, labs, and topics across Lattice.
PROMISE-AD, a novel survival analysis framework, was developed to predict individualized Alzheimer's Disease (AD) progression by modeling conversion from cognitively normal (CN) to mild cognitive impairment (MCI) and from MCI to AD dementia. The model uses a temporal Transformer to fuse representations of longitudinal patient data, including standardized measurements, missingness masks, and time-normalized slopes, to estimate a progression score and latent discrete-time mixture hazards. Evaluated on the ADNI/TADPOLE datasets, PROMISE-AD achieved state-of-the-art performance in predicting CN-to-MCI conversion (IBS 0.085, C-index 0.808) and strong performance in MCI-to-AD conversion (C-index 0.894), demonstrating the effectiveness of progression-aware survival modeling for AD risk estimation.
Predicting Alzheimer's progression with high accuracy just got a boost: a new model leverages longitudinal patient data and Transformers to estimate conversion risks, outperforming existing methods in key metrics.
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.
