Classification of Hypertension Disease Using Backpropagation Algorithm Based on Medical Record Data

Hypertension is a medical condition that can increase the risk of cardiovascular disease and death, and is often referred to as the “silent killer” due to its invisible symptoms. This study aims to develop a hypertension classification model using the Backpropagation Neural Network (BPNN) algorithm by considering risk factors such as smoking habits and family history of hypertension. The secondary data used comes from Dumai City Hospital with a total sample of 974 medical records which include attributes of age, gender, family history of hypertension, systolic blood pressure, diastolic blood pressure, body weight, and smoking status. The research process includes data integration, preprocessing (cleaning, transformation, and normalization), and division of training and test data. The model was tested with various hyperparameter configurations such as the number of neurons, activation function, optimizer, and learning rate. The best combination is obtained in the configuration of Tanh, Adamax, learning rate 0.01, and 50 hidden neurons with data division of 80:20 and 90:10. The loss graph of the model in the 80:20 scheme shows a low loss value and a steady decline with a small difference between training and validation loss, better than the 90:10 scheme. In this configuration, the BPNN model produces 100% testing accuracy, higher than previous models such as C4.5 Decision Tree and Random Forest. Smoking habits and family history of hypertension were found to have a significant influence on the risk of hypertension. With high accuracy and good model stability, the results of this study open up opportunities for the development of artificial intelligence-based prediction systems for early diagnosis and more effective management of hypertension.