Real-time Physiological Monitoring Interactive Model for Chronic Diseases in the Elderly

Xiaoyan Xu^*Xia Chen

• Feicheng Hospital affiliated to Mount Taishan Medical College, Feicheng, China

With the acceleration of aging process, the incidence of chronic diseases continues to rise, and the elderly population has an increasing demand for precise and convenient health monitoring. To address the physiological and mental state assessment issues faced in chronic disease monitoring for elderly people living at home, this study utilizes a multi-head self attention mechanism to detect arrhythmias. Firstly, the db6 wavelet transform is utilized to denoise and improve the quality of the electrocardiogram signal. Secondly, semantic features are extracted through a linear projection layer, and spatiotemporal representation is achieved by combining position encoding. A multi-head self-attention mechanism is adopted to capture the global semantic connections of signal segments and achieve the integration of non adjacent information. For mental state recognition, the system uses multi-channel convolutional attention mechanism to resample physiological signals and Butterworth filter to coordinate different sensor data, and uses multiple convolution kernels and maximum pooling operation to extract local features to guarantee the accuracy of pressure recognition. The results proved that the classification accuracy in arrhythmia detection reached 96.5%, significantly better than the conventional machine learning method's 89.2%. For mental state recognition, the model had an accuracy of 93.4% and demonstrated stable predictive ability in long-term real-time monitoring. In addition, the monitoring system integrated multiple sensors to achieve automatic collection, visualization, and health warning functions of physiological data. The proposed system had high accuracy and applicability. The research method has improved the monitoring efficiency of chronic diseases in the elderly, which can effectively support personalized management and intervention, and further provide reference for the optimization of intelligent health monitoring systems.