The OTNmanba-BiLSTM model provided high-precision predictions of exercise-induced improvements in cardiovascular function compared to conventional methods.
A novel OTNmanba-BiLSTM machine learning model utilizing wearable device data provides high-precision predictions of exercise-induced cardiovascular health improvements, outperforming conventional deep learning methods.
Tasa de eventos absoluta: 0% vs 0%
Cardiovascular diseases are among the leading chronic health conditions worldwide, with prolonged physical inactivity or improper exercise increasing the risk of their occurrence. Existing studies suggest that well-structured exercise interventions can significantly enhance cardiac function and improve the adaptability of the cardiovascular system. However, accurately predicting the long-term impact of different exercise durations and intensities on heart health remains challenging due to individual physiological differences, exercise habits, and lifestyle factors. This study employs a data-driven approach to develop a predictive model based on wearable device motion monitoring data, aiming to elucidate the dynamic relationship between exercise and cardiovascular health and support personalized exercise interventions. Volunteers wore wearable devices to continuously record multiple physiological parameters (e.g., heart rate, movement speed, and step frequency) during exercise, generating a time-series dataset. A bidirectional long short-term memory network (BiLSTM) was utilized for time-series prediction, integrated with manba and OrthoNets attention mechanisms, leading to the development of the OTNmanba-BiLSTM prediction algorithm. This hybrid model enhances forecasting accuracy and stability by capturing temporal dependencies and critical features in physiological data, enabling the quantification of exercise-induced effects on cardiovascular function. Experimental results demonstrate that the OTNmanba-BiLSTM model effectively captures the impact of exercise on cardiovascular function, providing high-precision predictions across different exercise durations and intensities. Compared to conventional deep learning methods, the proposed model exhibits superior long-term prediction accuracy and generalization capability. The findings validate the quantitative association between exercise interventions and cardiovascular health improvements, offering actionable insights for personalized exercise regimens. This study validates the effectiveness of data-driven modeling in understanding the exercise-cardiovascular health relationship and establishes the OTNmanba-BiLSTM algorithm as a robust tool for predicting long-term cardiac health outcomes. The framework provides a foundation for scalable applications in exercise rehabilitation and cardiovascular health management. Future research should prioritize optimizing model architectures, integrating individualized factors (e.g., genetics, clinical data), and expanding applications in chronic disease exercise prescription, thereby advancing the integration of precision medicine and sports science.
Li et al. (Thu,) reported a other. The OTNmanba-BiLSTM model provided high-precision predictions of exercise-induced improvements in cardiovascular function compared to conventional methods.