Open AccessBlood Glucose Prediction Model for Type 1 Diabetes based on Extreme Gradient Boosting
Author(s) -
Ganjar Alfian,
Muhammad Syafrudin,
Jongtae Rhee,
Muhammad Anshari,
Muhammad Rizki Darmawan Mustakim,
Imam Fahrurrozi
Publication year - 2020
Publication title -
iop conference series. materials science and engineering
Language(s) - English
Resource type - Journals
eISSN - 1757-899X
pISSN - 1757-8981
DOI - 10.1088/1757-899x/803/1/012012
Subject(s) - boosting (machine learning) , gradient boosting , mean squared error , predictive modelling , mathematics , mean squared prediction error , diabetes mellitus , artificial intelligence , statistics , medicine , computer science , endocrinology , random forest
Predicting future blood glucose (BG) level for diabetic patients will help them to avoid critical conditions in the future. This study proposed Extreme Gradient Boosting (XGBoost), an ensemble learning model to predict the future blood glucose value of diabetic patients. The clinical dataset of Type 1 Diabetes (T1D) patients was utilized and the prediction models were generated to predict future BG of 30 and 60 minutes ahead of time. The prediction models have been tested tofive children who develop T1D and showed that BG prediction model based on XGBoost outperformed other models, with average of Root Mean Square Error (RMSE) are 23.219 mg/dL and 35.800 mg/dL for prediction horizon (PH) 30 and 60 minutes respectively. In addition, the result showed that by utilizing statistical-based features as additional attributes, most of the performance of predictions model were increased.