Impact of sensor and measurement timing errors on model-based insulin sensitivity

Area of Science:

• Biomedical Engineering
• Clinical Informatics
• Endocrinology

Background:

• Insulin sensitivity (SI) parameter is crucial for glucose-insulin system models.
• Blood glucose (BG) sensor and measurement timing errors can affect SI identification.
• Impact on glycaemic variability and control analyses is a concern.

Purpose of the Study:

• Assess the impact of BG sensor and measurement timing errors on SI.
• Evaluate the hour-to-hour variability of identified SI.
• Investigate implications for glycaemic control protocols.

Main Methods:

• Retrospective analysis of clinical data from 270 ICU patients.
• Development of error models for glucometer and measurement timing.
• Monte-Carlo analysis to investigate the effect of errors on SI.

Main Results:

• Timing errors alone showed little clinical impact on SI level or variability.
• Combined sensor and timing errors significantly impacted identified SI levels.
• Identified SI values were within 29% of true values (90th percentile) under combined errors.

Conclusions:

• Timing errors have minor effects, but combined sensor/timing errors are significant for glycaemic control.
• Physiological factors in critically ill patients may overshadow measurement errors.
• Glycaemic control protocols must be robust to glucometer errors and avoid aggressive insulin dosing.