Prediction of Hypertension using an Android Application and Feature Selection Scheme
Prediction of diseases in the earlier stages helps an individual to improve his health and to avoid the dangerous health situations. This paper proposes a clinical decision support system, to provide an earlier prediction of hypertension based on the risk factors of an individual. The proposed system consists of the univariate selection method and the feature importance method to select the features that have the strongest relationship with the output variable i.e., disease prediction variable and random forest classifier method to predict the disease. Three datasets were utilized to extract the risk factors related to the disease hypertension. The most significant risk factors were determined from the dataset using the combination of both the univariate selection method and the feature importance method. Accuracy in the prediction of the disease is higher than the results obtained from using the methods separately. A mobile application has also been developed to provide a practical application of this proposed system. The risk factor data gathered from the mobile application is sent to the remote server where the machine learning algorithm is used to predict the outcome. The result is then immediately sent back to the application, where the result is displayed to the individual. Thus, it provides an effective way to find the risk of hypertension at an earlier stage.