Personalized Forecasting of Glycemic Control in Type 1 and 2 Diabetes Using Foundational AI and Machine Learning Models

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Cichosz, Simon Lebech, Hangaard, Stine, Kronborg, Thomas, Vestergaard, Peter, Jensen, Morten Hasselstrøm
Format: Preprint
Veröffentlicht: 2026
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866909979949137920
author Cichosz, Simon Lebech
Hangaard, Stine
Kronborg, Thomas
Vestergaard, Peter
Jensen, Morten Hasselstrøm
author_facet Cichosz, Simon Lebech
Hangaard, Stine
Kronborg, Thomas
Vestergaard, Peter
Jensen, Morten Hasselstrøm
contents Background: Accurate week-ahead forecasts of continuous glucose monitoring (CGM) derived metrics could enable proactive diabetes management, but relative performance of modern tabular learning approaches is incompletely defined. Methods: We trained and internally validated four regression models (CatBoost, XGBoost, AutoGluon, tabPFN) to predict six weekahead CGM metrics (TIR, TITR, TAR, TBR, CV, MAGE, and related quantiles) using 4,622 case-weeks from two cohorts (T1DM n=3,389; T2DM n=1,233). Performance was assessed with mean absolute error (MAE) and mean absolute relative difference (MARD); quantile classification was summarized via confusion-matrix heatmaps. Results: Across T1DM and T2DM, all models produced broadly comparable performance for most targets. For T1DM, MARD for TIR, TITR, TAR and MAGE ranged 8.5 to 16.5% while TBR showed large MARD (mean ~48%) despite low MAE. AutoGluon and tabPFN showed lower MAE than XGBoost for several targets (e.g., TITR: p<0.01; TAR/TBR: p<0.05 to 0.01). For T2DM MARD ranged 7.8 to 23.9% and TBR relative error was ~78%; tabPFN outperformed other models for TIR (p<0.01), and AutoGluon/ tabPFN outperformed CatBoost/XGBoost on TAR (p<0.05). Inference time per 1,000 cases varied markedly (PFN 699 s; AG 2.7 s; CatBoost 0.04 s, XGBoost 0.04 s). Conclusions: Week-ahead CGM metrics are predictable with reasonable accuracy using modern tabular models, but low-prevalence hypoglycemia remains difficult to predict in relative terms. Advanced AutoML and foundation models yield modest accuracy gains at substantially higher computational cost.
format Preprint
id arxiv_https___arxiv_org_abs_2601_00613
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Personalized Forecasting of Glycemic Control in Type 1 and 2 Diabetes Using Foundational AI and Machine Learning Models
Cichosz, Simon Lebech
Hangaard, Stine
Kronborg, Thomas
Vestergaard, Peter
Jensen, Morten Hasselstrøm
Other Quantitative Biology
Background: Accurate week-ahead forecasts of continuous glucose monitoring (CGM) derived metrics could enable proactive diabetes management, but relative performance of modern tabular learning approaches is incompletely defined. Methods: We trained and internally validated four regression models (CatBoost, XGBoost, AutoGluon, tabPFN) to predict six weekahead CGM metrics (TIR, TITR, TAR, TBR, CV, MAGE, and related quantiles) using 4,622 case-weeks from two cohorts (T1DM n=3,389; T2DM n=1,233). Performance was assessed with mean absolute error (MAE) and mean absolute relative difference (MARD); quantile classification was summarized via confusion-matrix heatmaps. Results: Across T1DM and T2DM, all models produced broadly comparable performance for most targets. For T1DM, MARD for TIR, TITR, TAR and MAGE ranged 8.5 to 16.5% while TBR showed large MARD (mean ~48%) despite low MAE. AutoGluon and tabPFN showed lower MAE than XGBoost for several targets (e.g., TITR: p<0.01; TAR/TBR: p<0.05 to 0.01). For T2DM MARD ranged 7.8 to 23.9% and TBR relative error was ~78%; tabPFN outperformed other models for TIR (p<0.01), and AutoGluon/ tabPFN outperformed CatBoost/XGBoost on TAR (p<0.05). Inference time per 1,000 cases varied markedly (PFN 699 s; AG 2.7 s; CatBoost 0.04 s, XGBoost 0.04 s). Conclusions: Week-ahead CGM metrics are predictable with reasonable accuracy using modern tabular models, but low-prevalence hypoglycemia remains difficult to predict in relative terms. Advanced AutoML and foundation models yield modest accuracy gains at substantially higher computational cost.
title Personalized Forecasting of Glycemic Control in Type 1 and 2 Diabetes Using Foundational AI and Machine Learning Models
topic Other Quantitative Biology
url https://arxiv.org/abs/2601.00613