Collaborating with intelligent health monitoring in the clinical care of elderly patients can enhance understanding of changes in their body functions, thereby improving their quality of life and preventing complications. This paper focuses on the user characteristics and needs of geriatric patients, builds an intelligent health monitoring system based on C/S architecture, and develops a health data collection process for geriatric patients using smart wearables. We use the VMD algorithm to reduce the noise in the physiological signal data of elderly patients, then input it into a time-sequence convolutional network to extract the corresponding ECG signal features. We then combine this data with the LSTM model to classify the ECG signals, enabling us to diagnose the health of elderly patients. We established a continuity of care program based on the intelligent health monitoring system and designed a comparison experiment to evaluate the impact of the application. The VMD algorithm can recognize the insignificant signal peaks between 1.5s and 5.5s, and the health diagnosis model has the highest classification accuracy of 99.24% for ECG beats. In the continuity of care model, the elderly patients’ physiological function score was 68.42±4.76, and their serum ALB level was 35.79±6.72 g/L, which was 17.42% higher than the control group’s level after the intervention. It also helped the elderly patients’ mental health. The intelligent health monitoring system can dynamically acquire the physiological characteristics of elderly patients in real time and generate visualization results to help medical staff customize personalized care plans.