A Hybrid Machine Learning–Driven, Explainable Framework for Competition-Aware and Risk-Sensitive Athlete Nutrition

Abstract

The paper introduces a Hybrid AI Systems-based, Explainable Artificial Intelligence (XAI) framework to competition conscious and risk conscious athlete nutrition as a part of a safety-critical Decision Support Systems architecture. The system combines the multimodal food knowledge, the modespecific context-sensitive reasoning and the hierarchical decision modeling to produce high-quality individualized dietary proposals based on competition dynamic constraints. RiskSensitive Modeling is added to implement the medical safety limits with underperforming readiness. The multimodal athlete nutrition datasets obtained through experimental assessment showed better predictive reliability with a 0.95 accuracy, F1-score of 0.93 and a lower Risk Violation rate at 3%. The framework has a low Expected Calibration Error of 0.021, or can be sure of probabilistic confidence in the recommendations. The explainability consistency stands at 0.92, which makes valid the consistent and transparent attribution of decisions. Feasibility in real-time is established and end-to-end latency is 52 ms on average. The findings support the efficiency of the integration of Hybrid AI Systems, Context-Aware Decision Systems, and Personalized Health Informatics to provide interpretable, risk-conscious, and competition-wise nutrition intelligence to high-performance athletics settings.

Ähnliche Arbeiten