Raised Blood Pressure and Sugar Detection and Prediction Models Using Machine Learning Algorithms: As A Screening Tool for Hypertension and Diabetes رسالة ماجستير

Abstract

High blood pressure and raised blood sugar are key risk factors for many chronic Non Communicable Diseases (NCDs). Raised blood sugar is a strong indicator of prediabetes or diabetes mellitus. Similarly, high blood pressure is considered a sign of hypertension, which is identified as a key risk factor for developing heart and cardiovascular diseases. Interestingly, hypertension and diabetes mellitus are the top common global NCDs affecting the adult population not only the elderly. Recently, the prevalence of diabetes and hypertension has been increasing at a faster rate, especially in developing countries. The primary concern associated with these diseases is the potential for serious health complications to occur if it is not diagnosed early or not managed properly, which may progress poorly and lead to disabilities. Therefore, timely detection and screening of diabetes and hypertension is considered a crucial factor in treating and controlling those diseases and averting their progression into severe health consequences. Population screening for high blood pressure and raised blood sugar aims to identify individuals at risk before symptoms appear, enabling timely intervention and potentially improved health outcomes. However, implementing large-scale screening programs can be expensive, requiring testing, follow-up, and management resources, potentially straining healthcare systems. Given the above facts, this study presents supervised machine learning models to detect and predict raised blood pressure and sugar health conditions. The proposed prediction models utilize the related risk factors that are shared by the two health conditions of high blood pressure and raised blood sugar. These common risk factors involve age, body mass index, eating habits, physical activity, history of other diseases, and fasting blood sugar, obtained from the dataset of the STEPwise study of NCDs risk factors, collected from adults in the Palestinian community. The NCDs risk factors gathered by the STEPS dataset were used as input for building the prediction models, which were trained using various types of supervised-learning classification algorithms including Random Forest, XGBoost, Decision Tree, and Multilayer Perceptron. Based on the experimental results, the proposed models performed the best predictive power by employing the Random Forest algorithm, yielding an accuracy of 98.05%, and 94.76% for the raised blood sugar and blood pressure detection models respectively. Additionally, the experimental results for the models implemented using the other classifiers are promising. The raised blood pressure and sugar detection models can be improved by incorporating multiple separate measurements of fasting blood sugar, and blood pressure that are taken on various days, so it can be used as a highly efficient and V accurate diagnosis tool for diabetes, and hypertension, not only for screening purposes. In addition, it can be extended for determining the classification of the fasting blood sugar whether it is normal, impaired, or raised, and classifying the type of blood pressure disorders