Optimizing blood glucose predictions in type 1 diabetes patients using a stacking ensemble approach

Introduction: The diabetes pandemic, including 828 million adults worldwide in 2022, would benefit from continued development of novel, effective and accurate blood glucose prediction systems. Using the DiaTrend dataset, this study used stacking machine learning optimized by Grey Wolf Optimizer to c...

Full description

Saved in:
Bibliographic Details
Main Authors: Vincent B. Liu, Laura Y. Sue, Oscar Madrid Padilla, Yingnian Wu
Format: Article
Language:English
Published: Elsevier 2025-06-01
Series:Endocrine and Metabolic Science
Subjects:
Online Access:http://www.sciencedirect.com/science/article/pii/S2666396125000391
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850246361290637312
author Vincent B. Liu
Laura Y. Sue
Oscar Madrid Padilla
Yingnian Wu
author_facet Vincent B. Liu
Laura Y. Sue
Oscar Madrid Padilla
Yingnian Wu
author_sort Vincent B. Liu
collection DOAJ
description Introduction: The diabetes pandemic, including 828 million adults worldwide in 2022, would benefit from continued development of novel, effective and accurate blood glucose prediction systems. Using the DiaTrend dataset, this study used stacking machine learning optimized by Grey Wolf Optimizer to construct and assess prediction models for blood glucose levels in type 1 diabetes patients. Methods: The DiaTrend dataset includes 27,561 days of continuous glucose monitoring and 8220 days of insulin pump data for 54 patients with type 1 diabetes. Grey Wolf optimization was used to tune and evaluate three machine learning algorithms – Random Forest, LSTM, GRU – for blood glucose predictions, whose predictions were then combined into an XGBoost stacking ensemble meta-learner. Results: This study looked at three baseline algorithms for predicting blood glucose levels. Machine learning models Random Forest, LSTM, and GRU served as baselines, with MAE, RMSE, and MARD values. GRU had the best predictive accuracy of the initial models. Grey Wolf optimization contributed to achieving optimal baseline model results. Stacking ensemble learning via XGBoost meta-learner (MAE = 10.65, RMSE = 14.59, MARD = 6.98) achieved higher performance than the baseline models. Conclusion: The GRU method with Grey Wolf optimization outperformed the other models with the lowest MAE, RMSE, and MARD, but the Stacked XGBoost model fared best. These findings emphasize the need to improve parameter selection with approaches such as Grey Wolf or stacking ensemble methods to achieve accurate blood glucose predictions. These prediction models can aid in the continued development of monitoring devices, and algorithms for these devices, which contain alert systems for impending abnormal blood glucose levels, allowing for timely diabetes self-management.
format Article
id doaj-art-855423b62843452987ee2b4ac4785f1a
institution OA Journals
issn 2666-3961
language English
publishDate 2025-06-01
publisher Elsevier
record_format Article
series Endocrine and Metabolic Science
spelling doaj-art-855423b62843452987ee2b4ac4785f1a2025-08-20T01:59:13ZengElsevierEndocrine and Metabolic Science2666-39612025-06-011810025310.1016/j.endmts.2025.100253Optimizing blood glucose predictions in type 1 diabetes patients using a stacking ensemble approachVincent B. Liu0Laura Y. Sue1Oscar Madrid Padilla2Yingnian Wu3Department of Statistics and Data Science, University of California, Los Angeles, Los Angeles, CA, United States of America; Corresponding author at: 8125 Math Sciences Bldg, Los Angeles, CA 90095, United States of America.Division of Endocrinology, Diabetes, and Metabolism, Department of Medicine, UCLA David Geffen School of Medicine, Los Angeles, CA, United States of AmericaDepartment of Statistics and Data Science, University of California, Los Angeles, Los Angeles, CA, United States of AmericaDepartment of Statistics and Data Science, University of California, Los Angeles, Los Angeles, CA, United States of AmericaIntroduction: The diabetes pandemic, including 828 million adults worldwide in 2022, would benefit from continued development of novel, effective and accurate blood glucose prediction systems. Using the DiaTrend dataset, this study used stacking machine learning optimized by Grey Wolf Optimizer to construct and assess prediction models for blood glucose levels in type 1 diabetes patients. Methods: The DiaTrend dataset includes 27,561 days of continuous glucose monitoring and 8220 days of insulin pump data for 54 patients with type 1 diabetes. Grey Wolf optimization was used to tune and evaluate three machine learning algorithms – Random Forest, LSTM, GRU – for blood glucose predictions, whose predictions were then combined into an XGBoost stacking ensemble meta-learner. Results: This study looked at three baseline algorithms for predicting blood glucose levels. Machine learning models Random Forest, LSTM, and GRU served as baselines, with MAE, RMSE, and MARD values. GRU had the best predictive accuracy of the initial models. Grey Wolf optimization contributed to achieving optimal baseline model results. Stacking ensemble learning via XGBoost meta-learner (MAE = 10.65, RMSE = 14.59, MARD = 6.98) achieved higher performance than the baseline models. Conclusion: The GRU method with Grey Wolf optimization outperformed the other models with the lowest MAE, RMSE, and MARD, but the Stacked XGBoost model fared best. These findings emphasize the need to improve parameter selection with approaches such as Grey Wolf or stacking ensemble methods to achieve accurate blood glucose predictions. These prediction models can aid in the continued development of monitoring devices, and algorithms for these devices, which contain alert systems for impending abnormal blood glucose levels, allowing for timely diabetes self-management.http://www.sciencedirect.com/science/article/pii/S2666396125000391DiabetesPredictionMachine learningStackingGrey wolfTime series
spellingShingle Vincent B. Liu
Laura Y. Sue
Oscar Madrid Padilla
Yingnian Wu
Optimizing blood glucose predictions in type 1 diabetes patients using a stacking ensemble approach
Endocrine and Metabolic Science
Diabetes
Prediction
Machine learning
Stacking
Grey wolf
Time series
title Optimizing blood glucose predictions in type 1 diabetes patients using a stacking ensemble approach
title_full Optimizing blood glucose predictions in type 1 diabetes patients using a stacking ensemble approach
title_fullStr Optimizing blood glucose predictions in type 1 diabetes patients using a stacking ensemble approach
title_full_unstemmed Optimizing blood glucose predictions in type 1 diabetes patients using a stacking ensemble approach
title_short Optimizing blood glucose predictions in type 1 diabetes patients using a stacking ensemble approach
title_sort optimizing blood glucose predictions in type 1 diabetes patients using a stacking ensemble approach
topic Diabetes
Prediction
Machine learning
Stacking
Grey wolf
Time series
url http://www.sciencedirect.com/science/article/pii/S2666396125000391
work_keys_str_mv AT vincentbliu optimizingbloodglucosepredictionsintype1diabetespatientsusingastackingensembleapproach
AT lauraysue optimizingbloodglucosepredictionsintype1diabetespatientsusingastackingensembleapproach
AT oscarmadridpadilla optimizingbloodglucosepredictionsintype1diabetespatientsusingastackingensembleapproach
AT yingnianwu optimizingbloodglucosepredictionsintype1diabetespatientsusingastackingensembleapproach