Accuracy Upgrade of BRFSS Dataset Subject Heart Disease Using the Hybrid Algorithm

Abstract

Heart disease is a specific abnormal condition that includes heart-related and bloodstream-related illnesses. It can range from heart rhythm issues to blood vessel disorders. Medical practitioners use a patient''s medical history and tests like blood pressure, blood sugar, or cholesterol to make a diagnosis of heart disease. The Behavioral Risk Factor Surveillance System is the greatest telephone survey program in the United States for learning about Americans'' health-related risk behaviors, chronic illnesses, and use of preventative drugs (BRFSS). We may use the data to identify a variety of illness-related traits and forecast an individual's likelihood of developing a specific condition. Following that, we compare and contrast the accuracy of the outcomes obtained using the methods described in the journal, earlier methods for previous projects, and hybrid algorithms. The objective of this research is to surpass the result of the journal reference which is the minimum accuracy of 78% and the maximum of 90%. Meanwhile the results of our Hybrid Algorithm methods are 90% for XGBoost+Randomize Search and 91% for the other algorithms. Based on the results, we can conclude that by combining the classification algorithm with the hyperparameter optimization method, we may enhance classification accuracy. Because the presence of hyperparameters makes the limit more optimal and optimal for classification, resulting in more accurate findings. It can be seen that the results obtained from our project are slightly higher compared to the Journal. The highest accuracy from the Journal is 90% while our result is 91%.