A correction method using a support vector machine to minimize hematocrit interference in blood glucose measurements.

Area of Science:

• Biomedical Engineering
• Clinical Chemistry
• Medical Diagnostics

Background:

• Point-of-care testing glucose meters are crucial for managing diabetes and critical care, but hematocrit (Hct) levels can significantly impact their accuracy.
• Existing glucose measurement methods are susceptible to interference from varying Hct levels, leading to potential inaccuracies.

Purpose of the Study:

• To develop and validate a novel algorithm for glucose calculation that compensates for hematocrit (Hct) variations.
• To enhance the accuracy of blood glucose measurements obtained from point-of-care testing devices.

Main Methods:

• Developed an adaptive calibration curve using linear fitting prediction and a support vector machine for glucose calculation.
• Implemented a hematocrit (Hct) compensation method to minimize Hct interference.
• Evaluated performance against ISO 15197:2013 standards, including bias, coefficient of variation, and error grid analysis.
• Utilized chronoamperometry to verify Hct effects and compare the proposed method.

Main Results:

• The Hct compensation method demonstrated an average bias of 5.69%, significantly lower than the 12.08% bias of chronoamperometry.
• The proposed method showed improved accuracy and reliability in glucose measurements across varying Hct levels.
• Chronoamperometry results confirmed that Hct variations directly influence glucose concentration readings.

Conclusions:

• The developed Hct compensation algorithm effectively minimizes the impact of hematocrit on glucose meter accuracy.
• This method offers a significant improvement over existing chronoamperometry techniques for point-of-care glucose monitoring.
• The findings support the clinical utility of this enhanced method for accurate blood glucose management.