Investigating Trustworthiness of Nonparametric Deep Survival Models for Alzheimer's Disease Progression Analysis

Researchers evaluated fairness and feature importance in deep survival models for predicting Alzheimer's disease progression, proposing two novel time-dependent concordance metrics to detect learned bias that could produce unreliable predictions across demographic groups.