Random Forest Based Blood Pressure Prediction Model from ECG And PPG Signal

Cardiovascular disease is the "first killer" endangering global public health. Blood pressure waveform contains rich information of cardiovascular system and can provide clinical guidance for treatment of cardiovascular diseases. Machine learning based blood pressure prediction methods have been widely studied and showed better performance in recent years. This study combined pulse transit time and electrocardiogram morphological features to predict blood pressure and used random forest regression method to build blood pressure prediction model. The proposed model was systematically evaluated and showed a good performance on evaluation metrics over a large dataset. This study is helpful for the wide application of noninvasive BP measurement model.