Abstract
With the increasing need for precision and personalization in athletic training, artificial intelligence (AI) offers powerful tools for monitoring, evaluating, and optimizing athlete performance. This study presents a Generalized Adaptive Backpropagation (GABP) neural network that combines Genetic Algorithms (GA) with Backpropagation (BP) to model physiological adaptation and guide data-driven training decisions. The proposed system is designed specifically for sports training contexts, emphasizing real-time feedback and individualized load management based on multivariate physiological inputs. The GABP model extends traditional backpropagation networks by incorporating genetic algorithms, which accelerate convergence, reduce the risk of local minima, and enhance predictive accuracy. By integrating diverse input modalities-such as physical performance indicators, training schedules, and biometric data from wearables, video analysis, and electromyography (EMG)-the system achieves high-precision evaluation of training effectiveness. In addition to detecting fatigue and performance plateaus, the model provides interpretable insights to refine training strategies at both individual and programmatic levels. Experimental validation on physiological datasets demonstrates measurable improvements in prediction accuracy and convergence efficiency. This work contributes a domain-specific, interpretable, and adaptive modeling framework for modern sports science, supporting personalized coaching and intelligent performance monitoring.