Parallel Adaptive Multi-Objective Evolutionary Learning of Discretized Bayesian Network Classifiers for Clinical Data

In three linesBaymex, a multi-objective evolutionary algorithm, learns discretized Bayesian networks for clinical classification. Parallelized on 16 cores (54× speedup), it optimizes cross-entropy and BIC complexity. On real datasets (RADCURE, SUPPORT), it matches or outperforms decision trees, logistic regression, and random forests while producing interpretable models.

Summary generated by Claude — human-verified