Early Prediction of Type 2 Diabetes Using Machine Learning

Diabetes is a dangerous disease-causing serious impact on human health. Many people in the world lose their lives due to this disease. Till now more than <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$2-5$</tex> million people have lost their lives due to this severe disease. The increasing ratio of diabetes is showing that in 2040642 millions of people will be affected from it. There are also such patients who lose their lives just because they don't come to know that disease at proper time. Instant treatment is necessary to prevent it. For research purposes we take dataset from Kaggle diabetes dataset 2019. It has 953 instances and 17 attributes. Diabetes dataset 2019 contains labels as “Yes” and “No”. ML algorithms like Naïve Bayes, Decision tree, Random Forest and K-NN are applied. Algorithms accuracy are decision Tree (<tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{9 2. 3 1 \%}$</tex>), Naïve Bayes (86.01%), Random Forest(94.41%) and K-NN (87.06%). Algorithm providing highest accuracy is Random Forest which is <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{9 4. 4 2 \%}$</tex>. In the application of algorithm, we don't use SMOTE. The current type-2 diabetes prediction model is enough to help doctors to predict diabetes. However, as the time passes there can be many advancements in it.