Machine Learning Enabled Potential Approach for Cardiovascular Risks Prediction

Cardiovascular diseases (CVDs) are one of the leading causes of death worldwide, which accounts for almost 30% of total deaths. The major deaths can be curbed via early detection of the risk factors of CVDs. Regular monitoring of the cardiovascular parameters is one of the feasible solutions; however, due to limitations of lack of information in real-time recording and instability in the constant acquisition of the cardiovascular signals in the existing technologies of monitoring, it seems not to be possible. We have introduced a highly sensitive flexible piezoelectric sensor fabricated from a simple solution processable technique. Which consists of a highly sensitive, flexible, conformable piezoelectric film of fluoride-based polymers. It is perfectly working for detecting the physiological parameters of the body such as arterial pulse, carotid pulse, and gait analysis; since it is prompt responsive in the subtle pressure range (0.001-1 kPa), which is tested for assessing risk factors of cardiovascular diseases based on arterial pulse data. The recorded data is further fed into the machine learning algorithms, which learn from the data and recognize if an anomaly is detected in the arterial pulse based on the analysis of the pattern, assisting in early predicting the disease based on its data learning. We have developed numerous machine learning algorithms for training the sensor, such as pattern recognition and random forest, which gives classification prediction accuracy of ~94% and 98%, respectively.