In this paper, we propose a health monitoring system, which leverages the Internet of Things (IoT) in conjunction with time-series deep learning techniques. The main purpose is to build the system for connecting to the MiBand smart watch device as well as to provide timely notifications about health abnormal signs of the wearer to their relatives, doctors to assist in timely handling. Cardiovascular disease is one of the dangerous diseases, causing patients to have a very high risk of death and rapid death. Our proposed system can monitor, predict and detect abnormal heart rhythms, so it plays an important role for early detection of cardiac dysfunction, timely treatment, and reducing the risk of death. To solve this problem, we have developed a system to collect data from the MiBand device, build a model for predicting heart rate and an abnormal alert system to relatives and doctors. The heart rate prediction module is trained on six popular deep learning models. The experimental results show that all models have the mean absolute error (MAE) in the range of 3.4 to 4.2. The results of this study can be the basis for further studies to develop other health monitoring initiatives with comparable objectives.