Artificial Intelligence for Blood Glucose Level Prediction in Type 1 Diabetes: Methods, Evaluation, and Emerging...

Area of Science:

• Biomedical Engineering
• Artificial Intelligence in Healthcare
• Diabetes Technology

Background:

• Diabetes management requires precise blood glucose level (BGL) monitoring.
• Predicting BGL can prevent acute and chronic complications in type 1 diabetes mellitus (T1DM).
• Data-driven approaches offer potential for improved BGL prediction.

Purpose of the Study:

• To review data-driven methods for BGL prediction in T1DM.
• To explore the application of artificial intelligence (AI) techniques in BGL prediction.
• To summarize key aspects of developing and evaluating BGL prediction models.

Main Methods:

• Review of data-driven prediction models for BGL.
• Examination of model strategy, input, prediction horizon, and performance.
• Analysis of AI techniques including deep learning, transfer learning, ensemble learning, and causal analysis.

Main Results:

• Machine learning approaches frequently outperform classical time-series models for BGL prediction.
• Multivariate inputs and advanced AI methods enhance prediction accuracy.
• Valid statistical comparisons are crucial due to inter-individual variability in T1DM.

Conclusions:

• AI techniques show significant potential for improving BGL prediction accuracy in T1DM.
• Further research is needed for integrating affecting variables into prediction models.
• Continued exploration of AI strategies is vital for advancing BGL prediction.