This noninvasive blood glucose monitor measures concentrations of serum glucose with clinically acceptable accuracy and avoids individual daily calibration. A group of investigators led by Dr Carl Malchoff of the University of Connecticut Health Center in the USA, evaluated an advanced prototype of a blood glucose monitor. The monitor is a handheld, noninvasive device that uses principles of thermal emission spectroscopy. The researchers calibrated the prototype using 432 paired measurements of concentrations of serum glucose from 20 patients with diabetes who required insulin. The monitor was then evaluated blindly in 126 paired measurements from six patients with diabetes who required insulin. Measurements were also compared with laboratory reference values using an error-in-variables method.
The investigators showed that the measurements taken with the blood glucose monitor were reproducible. For the six patients studied, the standard deviation in measurements was 32mg/dL, the percent mean absolute relative error (%MARE) was 11.6 and the correlation coefficient was 0.87. For all measurements, the standard deviation was 27mg/dL, the %MARE was 8.6 and the correlation coefficient was 0.94. A Clark Error Grid analysis showed that all the measurements fell within zones A and B, with 90% falling within zone A. The investigators conclude that this noninvasive blood glucose monitor measures concentrations of serum glucose with clinically acceptable accuracy and avoids individual daily calibration. Diabetes Care 2002;25:2268-75
Diabetes In Control. A free weekly diabetes newsletter for Medical Professionals. News and information for Medical Professionals.