Brief

Researchers have developed LERD, a new Bayesian system designed to analyze multichannel EEG data for neurodegenerative classification, specifically targeting Alzheimer's disease. Unlike previous black-box methods, LERD explicitly models latent neural events and their relational dynamics without requiring pre-annotated events. The system combines continuous-time event inference with a stochastic generation process and an electrophysiology-inspired prior, offering theoretical guarantees for stability and tractability. Experiments on synthetic and real-world data show LERD outperforms existing approaches, providing physiologically aligned summaries that highlight group-level dynamical differences. AI