Generalizing CGM Sensor-Based Glucose Prediction across Age Cohorts Using LSTM Models
An In Silico Study with the UVA/Padova T1DMS Simulator
DOI:
https://doi.org/10.3991/ijoe.v22i04.60113Keywords:
Continuous Glucose Monitoring, Long Short-Term Memory, Type 1 Diabetes, Glucose Prediction, Closed-Loop Insulin SystemsAbstract
Accurate glucose prediction is necessary in enhancing insulin therapy and preventing harmful blood-sugar spikes in individuals with type 1 diabetes (T1D). Although deep learning models seem promising for glucose forecasting, it is unclear how they perform across various age groups that exhibit different metabolism profiles. This paper compares the performance of the long short-term memory (LSTM) models across age groups using simulated data from the UVA/Padova T1D Metabolic Simulator (T1DMS). Cohort-specific models achieved high withincohort performance (MAE < 2 mg/dL, r > 0.99, R2 > 0.99), indicating precise modeling of glucose–insulin dynamics within each group. Nevertheless, predictive accuracy decreased when models were applied across cohorts, especially when LSTM networks trained on adults were tested on younger groups, demonstrating physiological variability between ages in insulin sensitivity and glucose kinetics. Models trained on younger groups performed better on older populations, suggesting that a broader range of metabolic variation underlies increased adaptability. This study is the first to use the FDA-approved UVA/Padova T1DMS simulator to systematically assess age-dependent generalization in LSTM-based glucose prediction, offering a unique reproducible framework for developing adaptive e-health and closedloop insulin systems. Incorporating age-relevant physiological heterogeneity and adaptive modeling paradigms could help develop stronger, patient-specific glucose forecasting systems for safer and more effective diabetes management.
References
[1] NCD Risk Factor Collaboration (NCD-RisC), “Worldwide trends in diabetes since 1980: A pooled analysis of 751 population-based studies with 4.4 million participants,” Lancet, vol. 387, pp. 1513–1530, 2016.
[2] Emerging Risk Factors Collaboration, N. Sarwar, P. Gao, S. R. Seshasai, R. Gobin, S. Kaptoge, et al., “Diabetes mellitus, fasting blood glucose concentration, and risk of vascular disease: A collaborative meta-analysis of 102 prospective studies,” Lancet, vol. 375, pp. 2215–2222, 2010.
[3] J. M. Forbes and M. E. Cooper, “Mechanisms of diabetic complications,” Physiol. Rev., vol. 93, pp. 137–188, 2013.
[4] D. M. Nathan, “Long-term complications of diabetes mellitus,” N. Engl. J. Med., vol. 328, pp. 1676–1685, 1993.
[5] American Diabetes Association Professional Practice Committee, “Glycemic targets: Standards of medical care in diabetes—2022,” Diabetes Care, vol. 45, pp. S83–S96, 2022.
[6] D. Daneman, “Type 1 diabetes,” Lancet, vol. 367, pp. 847–858, 2006.
[7] L. Y. Melendez-Ramirez, R. J. Richards, and W. T. Cefalu, “Complications of type 1 diabetes,” Endocrinol. Metab. Clin. N. Am., vol. 39, pp. 625–640, 2010.
[8] R. J. McCrimmon and R. S. Sherwin, “Hypoglycemia in type 1 diabetes,” Diabetes, vol. 59, pp. 2333–2339, 2010.
[9] G. Desoye, D. Simmons, and M. van Poppel, “Identifying type 2 diabetes risk at birth: A new frontier for early prevention,” Diabetes Care, vol. 48, pp. 1320–1322, 2025.
[10] R. Taylor, “Understanding the cause of type 2 diabetes,” Lancet Diabetes Endocrinol., vol. 12, pp. 664–673, 2024.
[11] G. W. Ray et al., “Genetic and inflammatory factors underlying gestational diabetes mellitus: A review,” Front. Endocrinol., vol. 15, p. 1399694, 2024.
[12] N. Wang et al., “Risk factors screening for gestational diabetes mellitus heterogeneity in Chinese pregnant women: A case–control study,” Diabetes Metab. Syndr. Obes., vol. 14, pp. 951–961, 2021.
[13] J. Li et al., “Effect of gestational diabetes mellitus on pregnancy outcomes among younger and older women and its additive interaction with advanced maternal age,” Front. Endocrinol., vol. 14, p. 1158969, 2023.
[14] International Diabetes Federation, IDF Diabetes Atlas, 11th ed. Brussels, Belgium: International Diabetes Federation, 2025. [Online]. Available: https://www.diabetesatlas.org. [Accessed: Sept. 15, 2025].
[15] S. K. Vashist, “Continuous glucose monitoring systems: A review,” Diagnostics, vol. 3, pp. 385–412, 2013.
[16] V. Jaiswal, A. Negi, and T. Pal, “A review on current advances in machine learning based diabetes prediction,” Prim. Care Diabetes, vol. 15, pp. 435–443, 2021.
[17] M. Rahman, D. Islam, R. J. Mukti, and I. Saha, “A deep learning approach based on convolutional LSTM for detecting diabetes,” Comput. Biol. Chem., vol. 88, p. 107329, 2020.
[18] S. A. Alex et al., “Deep LSTM model for diabetes prediction with class balancing by SMOTE,” Electronics, vol. 11, p. 2737, 2022.
[19] S. Soltanizadeh and S. S. Naghibi, “Hybrid CNN–LSTM for predicting diabetes: A review,” Curr. Diabetes Rev., vol. 20, pp. 77–84, 2024.
[20] P. Dharanyadevi et al., “Comparative analysis of SVM and LSTM models for predicting diabetes: Evaluating performance and interpretability,” in Integrated Technologies in Electrical, Electronics and Biotechnology Engineering. Boca Raton, FL, USA: CRC Press, 2025, pp. 294–299.
[21] F. Mohsen et al., “A scoping review of artificial intelligence-based methods for diabetes risk prediction,” NPJ Digit. Med., vol. 6, p. 197, 2023.
[22] B. P. Kovatchev, M. Breton, C. Dalla Man, and C. Cobelli, “In silico preclinical trials: A proof of concept in closed-loop control of type 1 diabetes,” J. Diabetes Sci. Technol., vol. 3, pp. 44–55, 2009.
[23] The Epsilon Group, “Diabetes simulation,” 2025. [Online]. Available: https://tegvirginia.com/services/diabetes-simulation/. [Accessed: Oct. 15, 2025].
[24] W. Wang and Y. Lu, “Analysis of the mean absolute error (MAE) and the root mean square error (RMSE) in assessing rounding model,” IOP Conf. Ser. Mater. Sci. Eng., vol. 324, p. 012049, 2018.
[25] T. O. Hodson, “Root-mean-square error (RMSE) or mean absolute error (MAE): When to use them or not,” Geosci. Model Dev., vol. 15, pp. 5481–5487, 2022.
[26] J. Freiburghaus, A. Rizzotti, and F. Albertetti, “A deep learning approach for blood glucose prediction of type 1 diabetes,” in Proc. 5th Int. Workshop Knowl. Discovery Healthcare Data (KDH 2020), 2020, pp. 29–30.
[27] G. James, D. Witten, T. Hastie, and R. Tibshirani, An Introduction to Statistical Learning: With Applications in R. New York, NY, USA: Springer, 2013.
[28] E. T. Chen, J. H. Nichols, S. H. Duh, and G. Hortin, “Performance evaluation of blood glucose monitoring devices,” Diabetes Technol. Ther., vol. 5, pp. 749–768, 2003.
[29] D. Giavarina, “Understanding Bland–Altman analysis,” Biochem. Med., vol. 25, pp. 141–151, 2015.
[30] M. M. M. Mansour, “Non-parametric statistical test for testing exponentiality with applications in medical research,” Stat. Methods Med. Res., vol. 29, pp. 413–420, 2020.
[31] S. Thamilselvan and N. A. Ahad, “Comparison of post hoc tests on non-communicable diseases data,” Quantum J. Eng. Sci. Technol., vol. 4, p. 12, 2023.
[32] R. A. Williams et al., “Predictors of postprandial glycaemia, insulinaemia and insulin resistance in adolescents,” Br. J. Nutr., vol. 125, pp. 1101–1110, 2021.
[33] D. Montt-Blanchard, R. Sánchez, K. Dubois-Camacho, J. Leppe, and M. T. Onetto, “Hypoglycemia and glycemic variability of people with type 1 diabetes with lower and higher physical activity loads in free-living conditions using continuous subcutaneous insulin infusion with predictive low-glucose suspend system,” BMJ Open Diabetes Res. Care, vol. 11, p. e003082, 2023.
[34] P. Tsarkova et al., “CSII is related to more stable glycemia in adults with type 1 diabetes,” Endocrine, vol. 75, pp. 776–780, 2022.
[35] A. Khadilkar and C. Oza, “Glycaemic control in youth and young adults: Challenges and solutions,” Diabetes Metab. Syndr. Obes., vol. 15, pp. 121–129, 2022.
Downloads
Published
How to Cite
Issue
Section
License
Copyright (c) 2026 Saleh I. Alzahrani
This work is licensed under a Creative Commons Attribution 4.0 International License.
