A Non-Contact Heart Rate Measurement System Based On LSTM Neural Network

The proposed method combines a non-contact measuring device with an LSTM neural network to provide a simple and precise wireless system to measure heart rate. The movement of the thoracic surface is modeled from the effects of breathing motion and cardiac contractility. Then, we generate a dataset for training the model. Finally, the trained model is used to derive heart rate estimates from medical radar data. These estimates are evaluated through Root Mean Square Error and Heart Rate Error Average.