Predicting The Type of Treatment for Diabetes Patients using Machine Learning Algorithms

Prediction of treatment type for the patients of diabetes via Machine Learning (ML) techniques is one of health care field. It aims to analyze and evaluate the personal data of patients for determining the optimum treatment. This field can provide an effective and reliable method to improve diabetes management by reducing the need of trials and errors in determining appropriate treatment. This paper targets to predict the appropriate treatment for diabetic patients using ML techniques. The study addresses the challenges of determining whether the injections or the medications are suitable for individual patients by analyzing personal health data. The dataset, which will be used in this study, was collected from a specialized diabetes and endocrinology center in Misurata state in Libya. This dataset includes key features like Body Mass Index, blood pressure, levels, HbA1c levels and blood sugar. The methodology includes several stages: data collection through structured questionnaires, data preprocessing (dealing with missing data, normalization, feature selection, class imbalance treatment using SMOTE), and application of ML algorithms. The study uses Support Vector Machine, Random Forest, Decision Tree, and Gradient Boosting algorithms. Two sets of experiments were performed: the first, was done by using default parameters and the second by using hyperparameter tuning via Grid Search to improve model’s performance. Preliminary results showed that the Gradient Boosting algorithm achieved the highest accuracy of 74%. However, after hyperparameter tuning, the Random Forest algorithm showed superior performance with 89% accuracy rate, followed by 85% Gradient Boosting. The Support Vector Machine and Decision Tree algorithms also showed significant improvements in accuracy after the optimization. The results highlight the effectiveness of ML models in predicting diabetes treatment.