LSTM-Based Heart Rate Dynamics Prediction During Aerobic Exercise for Elderly Adults

This paper presents a specialized LSTM-based framework for predicting heart rate dynamics during aerobic exercise in elderly adults. The proposed model incorporates temporal feature extraction mechanisms and adaptive learning strategies specifically designed for age-related cardiovascular characteristics. A comprehensive dataset comprising 2,484,000 data points collected from 230 elderly participants (aged 65-80 years) during structured aerobic exercise sessions was utilized for model development and validation. The architecture employs a multi-stage preprocessing pipeline and specialized LSTM layers optimized for cardiovascular pattern recognition. Experimental results demonstrate superior prediction accuracy with a mean absolute error of 1.89 beats per minute, representing a 27.5% improvement over traditional methods. The model achieves real-time prediction capabilities with a latency of 28.5 milliseconds while maintaining a 94.3% signal quality preservation rate. Performance evaluation across multiple clinical settings validates the model's robustness and practical utility in elderly healthcare monitoring. The framework demonstrates significant potential for integration into existing cardiac monitoring systems, achieving a false positive rate of 2.1% and a false negative rate of 1.8%. The proposed approach establishes a foundation for enhanced cardiovascular risk assessment and personalized exercise prescription in elderly populations.