Using machine learning to predict hypertension from a clinical dataset
Daniel Lafrenière, Farhana H. Zulkernine, David Barber, Ken Martin · 2016
Hypertension is an illness that often leads to severe and life threatening diseases such as heart failure, thickening of the heart muscle, coronary artery disease, and other severe conditions if left untreated. An artificial neural network is a powerful machine learning technique that allows prediction of the presence of the disease in susceptible populations while removing the potential for human error. In this paper, we identify the important risk factors based on patients' current health conditions, medical records, and demographics. These factors are then used to predict the presence of hypertension in an individual. These risk factors are also indicative of the probability of a person developing hypertension in the future and can, therefore, be used as an early warning system. We present a neural network model for predicting hypertension with about 82% accuracy. This is good performance given our chosen risk factors as inputs and the large integrated data used for the study. Our network model utilizes very large sample sizes (185,371 patients and 193,656 controls) from the Canadian Primary Care Sentinel Surveillance Network (CPCSSN) data set. Finally, we present a literature study to show the use of these risk factors in other works along with experimental results obtained from our model.