Hypertension Disease Predictions with Various Models Using Data Science Framework

Hypertension is a common disease which may lead to incurable situation if it is not well-treated. By extracting knowledge from large datasets, data mining can be used for the hypertension prediction and diagnosis. In this research, three data mining techniques, the support vector machine (SVM), k-nearest neighbors (k-NN), and decision tree, were developed to predict the hypertension with two datasets. These techniques were implemented on Google Colab using Python codes. Experimental results illustrate that the accuracy and area under the receiver operating characteristic (ROC) curve performance of all these three techniques on first dataset is 81% and 80.64% for the SVM, 74% and 73.97% for the k-NN, and 74% and 73.86% for the decision tree, respectively. The accuracy and ROC curve performance of these three approaches on the second dataset were 73% and 72.48% for the SVM, 76% and 75.75% for the k-NN model, and 75% and 74.49% for the decision tree, respectively. The results indicate that the SVM, k-NN, and decision tree, are effective in predicting the hypertension based on the obtained important features of the patients, and the k-NN is among the best choice.