Heart Disease Prediction Using Interquartile Range Preprocessing and Hypertuned Machine Learning

Machine learning and data mining techniques have recently revolutionized detection of chronic diseases such as heart disease. Existing literature has established the potential of using machine learning algorithms to develop models to automate the detection of heart disease. Doing so will drastically reduce the overall cost of heart disease treatment. This will make treatment more accessible to the broader population in developing and underdeveloped countries. However, existing machine learning methods cannot offer the level of accuracy and cost effectiveness needed for real life implementations. In this study, we aimed to tackle these problems by proposing machine learning models that offer near perfect accuracy and performance. We applied several supervised hyper tuned machine learning classifiers, and low-cost neural networks and compared the accuracy and performance of these models. We achieved a peak accuracy of 99.02 percent with KNN. Thus, our study shows the viability of using simple machine learning with hypertuning for real life implementation of automated heart disease detection.