Intelligent generation and optimization of personalized physical training programs based on deep learning

The application of artificial intelligence (AI) in sports and health management has resulted in more precise and adaptive fitness programs. Traditional training plans often lack personalization and dynamic adjustment based on real-time physiological data, thereby reducing their effectiveness. This research proposes an intelligent framework for generating and optimizing personalized physical training programs using Tug War Optimized Intelligent Spiking Neural Network (TWO-ISNNet), which combines Tug War Optimization (TWO) with an intelligent Spiking Neural Network (ISNNet) to provide adaptive, real-time exercise guidance. Collect participant demographic, health, and wearable sensor data (heart rate, VO₂ max, steps, recovery time). Use normalization and missing value imputation for preprocessing. Principal Component Analysis (PCA) and Time-FrequencyAnalysis (TFA) are used to identify significant physiological and performance variables. Train the ISNNet on extracted features to identify temporal patterns in fitness response. Use TWO to fine-tune network parameters, increasing forecast accuracy and adaptability. The project is developed with Python and variations of the personalized exercise plans based on their intensity, duration, and type. When the proposed approach is used against the baseline models, the precision is 94.6%. The TWO-ISNNet approach generated adaptive exercise programs with high predictive accuracy and improved personalization efficiency based on physiological input data. The proposed solution focuses on the importance of the TWO and ISNNet combination to realize intelligent, real-time, and customized physical training, which offers a scalable computational framework for personalized fitness planning.