Post-prandial plasma glucose prediction in type I diabetes based on Impulse Response Models

Area of Science:

• Biomedical Engineering
• Computational Biology
• Endocrinology

Background:

• Accurate prediction of blood glucose fluctuations is crucial for diabetes management.
• Individualized glycemic response to meals and insulin varies significantly.
• Existing predictive models may lack precision for short-term forecasting.

Purpose of the Study:

• To develop and evaluate Finite Impulse Response (FIR) models for predicting glycemic responses.
• To assess the efficacy of short-term individualized predictors for 20- and 60-minute intervals.
• To compare the performance of developed predictors against established benchmarks.

Main Methods:

• Collected data from 18 patients regarding meal intake and rapid insulin administration.
• Estimated FIR models to capture the dynamic impact of meals and insulin on glucose levels.
• Developed and tested individualized short-term predictors based on the FIR models.
• Utilized Clarke Grid Analysis for evaluating prediction accuracy and safety.

Main Results:

• Predictive models demonstrated high accuracy, with over 94% and 75% of predictions falling within the clinically acceptable Zone A at 20 and 60 minutes, respectively.
• Error rates in the problematic C/D/E zones were minimal, below 1% and 3% at 20 and 60 minutes.
• The performance of the developed predictors was found to be competitive with existing published results.

Conclusions:

• FIR models provide a robust framework for understanding and predicting meal and insulin impacts on glucose.
• Individualized short-term glucose prediction models show significant promise for improving diabetes self-management.
• The developed predictors offer a competitive and accurate tool for real-time glycemic monitoring.