The Clinical Accuracy of Non-Invasive Glucose Monitoring for ex vivo Artificial Pancreas

Diabetes is a serious worldwide epidemic that affects a growing portion of the population. While the most common method for testing blood glucose levels involves finger pricking, it is painful and inconvenient for patients. We compared the glucose levels measured noninvasively from tears to those obtained from capillary blood in 10 diabetic and non-diabetic patients. In addition, we assessed the clinical accuracy of the newly developed non-invasive tear glucose monitoring system using a Clarke Error Grid Analysis (EGA) and tested if such a relationship is individualized or universal. A glucose meter circuit was built to analyze capillary blood and right- and left-eye tear fluids, and the results
were compared with those of a commercial glucose meter, which was used as the reference in the EGA. The prediction of glucose values from the voltages using linear regression showed the non-invasive system to be clinically accurate and universal. This result suggests that the coupling of a non-invasive glucose monitoring system, such as one that detects glucose in tears, with an insulin pump may be clinically feasible as an ex vivo artificial pancreas treatment. By employing this ex vivo artificial pancreas, many of the complications associated with frequent hyperglycemia and hypoglycemia events could be avoided, increasing the quality of life.