Feature Importance Analysis for Glucose Level Detection in Type 2 Diabetes using Machine Learning

Abstract

In this paper, we present a machine learning-based approach for the detection of glucose levels in type 2 diabetes patients. Our approach utilizes physiological parameters such as Body Mass Index (BMI), age, sex, and blood pressure, along with glucose levels, to train a predictive model. A dataset comprising demographic information, clinical history, and laboratory test results of 500 type 2 diabetes patients was collected for training and validation. Logistic regression, support vector machines, and random forest classifiers were trained and evaluated using various performance metrics, including accuracy, sensitivity, specificity, and area under the Receiver Operating Characteristic (ROC) curve. Results showed that the random forest classifier outperformed the other models, achieving an accuracy of 85% and an AUC-ROC score of 0.90. Feature importance analysis identified age, BMI, and blood pressure as the most critical predictors for glucose level detection in type 2 diabetes patients. Our proposed machine learning-based approach demonstrates promising results for the accurate detection of glucose levels in type 2 diabetes patients. It has the potential to assist healthcare professionals in making timely and accurate decisions regarding diabetes management. Furthermore, our findings provide valuable insights into the essential predictors for glucose level detection, which can guide future research in this area.