The feasibility of measuring blood glucose concentration in patients with diabetes mellitus using near infrared (NIR) spectroscopy and multivariate data analysis techniques has been established using general purpose, scientific-grade infrared spectrophotometers under the 90-049 UNM HRRC protocol. Now that the concept of painless, bloodless, and reagent-less clinical laboratory tests has been shown to be viable, the challenge is to bring this technology to the patients who need it. We must develop a device that is portable, affordable, and rugged - all while improving accuracy. We must also improve our understanding of the nature of glucose in whole, intact skin. Since the light beam cannot discriminate plasma glucose in skin blood vessels from glucose in the larger skin interstitial fluid space, the exact nature of the glucose in all tissue compartments contributing to the light-beam signal must be elucidated. The noninvasive measurement can be no more accurate than the reference measurement on which it was based. Finally, we have found that the accuracy of the noninvasive measurement also worsens as a function of time from the initial measurement. This effect is independent of instrumentation drift and is due to physical changes in the patient's skin such as variation of hydration, collagen, vascularity, glycation end-products, etc. To account for these sources of physiologic interference they must be evaluated independently and then compared to errors of the noninvasive predictions over time.
