Machine learning approach for predicting heart and diabetes diseases using data-driven analysis

Abstract

Environmental changes and food habits affect people's health with numerous diseases in today's life. Machine learning is a technique that plays a vital role in predicting diseases from collected data. The health sector has plenty of electronic medical data, which helps this technique to diagnose various diseases quickly and accurately. There has been an improvement in accuracy in medical data analysis as data continues to grow in the medical field. Doctors may have a hard time predicting symptoms accurately. This proposed work utilized Kaggle data to predict and diagnose heart and diabetic diseases. The diseases heart and diabetes are the foremost cause of higher death rates for people. The dataset contains target features for the diagnosis of heart disease. This work finds the target variable for diabetic disease by comparing the patient's blood sugars to normal levels. Blood pressure, body mass index (BMI), and other factors diagnose these diseases and disorders. This work justifies the filter method and principal component analysis for selecting and extracting the feature. The main aim of this work is to highlight the implementation of three ensemble techniques-Adaptive boost, Extreme Gradient boosting, and Gradient boosting-as well as the emphasis placed on the accuracy of the results.