Prediction of Hypertension using Machine Learning

Extreme and life-threatening illnesses caused by hypertension include coronary heart failure, thickening of the heart muscles, and different ailments if left untreated. Using a machine learning approach, we can predict the presence of the disease in a susceptible population of individuals. In this paper, a group of patients’ blood pressure measurements is acquired using an Ambulatory Blood Pressure Monitor (ABPM), and this data is utilized for training a machine learning model, which in turn learns the pattern of blood pressure levels categorized as hypertension. The model is trained and tested by feeding into it test data of a patient's 24-hour (ambulatory) blood pressure readings. Based on these readings, the model predicts the patient's blood pressure condition. This obtained prediction is then sent to the concerned medical professional via an automatically generated email. We present a machine learning model for predicting hypertension risk with approximately 0.0512 as the root mean square value and its obtained confusion matrix denoting its accuracy. With the chosen risk factors as inputs for the model combined with a large dataset used for training, the output produced can be concluded as a good performance.