Study of Short-Term Personalized Glucose Predictive Models on Type-1 Diabetic Children

TL;DR

This study compares various machine learning models, including a novel Gaussian Process approach, for short-term personalized glucose prediction in children with Type 1 Diabetes, highlighting their strengths and clinical relevance.

Contribution

It introduces a comparative analysis of multiple models, including a novel GP-DP kernel, for personalized glucose forecasting in pediatric diabetes management.

Findings

GP-DP achieved the lowest RMSE among models.

LSTM, SVR, and GP-DP showed acceptable clinical performance.

Models excelled in different glycemia regions.

Abstract

Research in diabetes, especially when it comes to building data-driven models to forecast future glucose values, is hindered by the sensitive nature of the data. Because researchers do not share the same data between studies, progress is hard to assess. This paper aims at comparing the most promising algorithms in the field, namely Feedforward Neural Networks (FFNN), Long Short-Term Memory (LSTM) Recurrent Neural Networks, Extreme Learning Machines (ELM), Support Vector Regression (SVR) and Gaussian Processes (GP). They are personalized and trained on a population of 10 virtual children from the Type 1 Diabetes Metabolic Simulator software to predict future glucose values at a prediction horizon of 30 minutes. The performances of the models are evaluated using the Root Mean Squared Error (RMSE) and the Continuous Glucose-Error Grid Analysis (CG-EGA). While most of the models end up having low RMSE, the GP model with a Dot-Product kernel (GP-DP), a novel usage in the context of glucose prediction, has the lowest. Despite having good RMSE values, we show that the models do not necessarily exhibit a good clinical acceptability, measured by the CG-EGA. Only the LSTM, SVR and GP-DP models have overall acceptable results, each of them performing best in one of the glycemia regions.