ML based Blood Glucose Level Detection using Microwave based sensors

This article presents a new, painless way to monitor blood sugar levels without using needles. It uses a microwave based antenna as a sensor to detect changes in blood glucose levels. In order to accurately predict the blood sugar level, machine learning models such as Random Forest, XGBoost, CatBoost, and Logistic Regression are trained using the antenna’s S-parameter changes. The study found that some models, especially CatBoost, work very well with an exceptional AUC of 0.97. Overall, this method that uses antenna as sensor for capturing S parameter data for varying blood glucose levels with Machine learning offer a safe, quick, and comfortable alternative for people with diabetes to regularly check their blood sugar levels.