Empirical Statistical Estimation of Glucose Concentration Using Thermal Radiation

OptiScan Biomedical Corporation's Non-Invasive Glucose Monitor has demonstrated the capability to measure patient blood glucose concentrations using an induced cyclic temperature variation that modulates the natural mid-infrared radiation coming from the patient's tissue just below the surface of the skin. ALPHATECH, Incorporated, can leverage the information extracted by this method using additional signal processing and nonlinear statistical modeling techniques. Glucose concentration affects IR absorption and hence the detected amplitude and phase at the thermal driving frequency. Currently, only the phase information is exploited in the Glucose Monitor. Modeling the signal amplitude and phase within a Support Vector Machine (SVM) framework, an initial simulation shows that the SVM predictive error is a factor of five smaller than that of a linear model. Both the linear and SVM model methods account for confounding variables in the simulation, and offer the potential for estimating the concentrations of other analytes of interest for metabolic monitoring and disease diagnosis. The ALPHATECH-OptiScan team proposes to demonstrate that advanced nonlinear methods are capable of accurately estimating glucose concentration levels in the presence of confounding metabolites in in vitro tests as the first step in the development of a breakthrough noninvasive metabolic monitoring device.