Hypertension prediction and analysis of risk factors based on physical examination report data

Hypertension is the most common chronic disease and the main risk factor for cardiovascular and cerebrovascular diseases. This paper presents a method to predict hypertension using Deep Neural Network(DNN) by physical examination report dataset, and compares the performance of DNN with the ensemble learning algorithm based on decision trees. The results indicate that the DNN demonstrates distinct advantages across various evaluation metrics, achieving an accuracy rate of 89.54%, a precision rate of 87.89%, a recall rate of 91.85% and the f1 score is 89.82%. Subsequently, Random Forest(RF) algorithm is selected to analyze the risk factors from the physical examination reports. The findings reveal that age, weight, triglyceride(TG), glutamic oxaloacetic transaminase(GOT), gamma-glutamyl transpeptidase(GGT), basophil(BASO), ca199, Glycated hemoglobin HbA1c(HbA1c), RBC average distribution width(RDW), Alanine aminotransferase(ALT) are among the top 10 risk factors for hypertension.