An efficient Blood pressure estimation algorithm based on machine learning using a mixture of non-contact hand-face PPG signals features

Hypertension, or high blood pressure, occurs when blood forcefully presses against artery walls, increasing heart and health risks. Blood pressure is typically measured with a sphygmomanometer, but this device cannot be used in some cases like burned bodies. To avoid this problem, a cuffless devices are used based on a non-contact signal called remote photo-plethysmography signal which is extracted from the face or the hand in some conditions using a special camera. In fact, the blood pressure is estimated based on the features extracted from this signal. In the presented work, we introduce a rapid algorithm for blood pressure estimation using a mixture of non-contact handface PPG signal features. In fact, the signals are collected from the face and the hand simultaneously. Then, they are used as an input of preprocessing blocks, which makes the feature extraction operation very useful and reliable. Finally, the machine-learning algorithm performs the estimate of blood pressure using the derived pulse time transit. Experimental results are very encouraging and interesting.