A Multimodal and Explainable Machine Learning Approach to Diagnosing Multi-Class Ejection Fraction from Electrocardiograms

An arXiv RSS batch (multiple new CS/ML submissions) covering diverse machine-learning topics: a multimodal, explainable XGBoost model for 4-way left ventricular ejection fraction (LVEF) classification from ECG + EHR with high AUROCs and SHAP explainability; a PDE energy-driven iterative solver avoiding matrix assembly; a survey of GNN-based communication in multi-agent RL; an information-theoretic re-formulation of KV-cache eviction (CapKV) for long-context LLM inference; identification of mini-batch composition bias in link-prediction training; an agenda-setting survey of open problems in ‘fr