Hannu Harjunmaa

Development of a non-invasive blood glucose monitor: application of artificial neural networks for signal processing

We have developed a noninvasive blood glucose measuring instrument, based on application of the Optical Bridge/sup TM/ in the near-infrared region. This exploratory research is an endeavor to evaluate the possibility of increasing the performance of the noninvasive glucose monitor by employing Artificial Neural Networks (ANN). The objective of this research is to design an ANN to interpret the instruments outputs as well as the system parameters, and correlate them with blood glucose levels. The main hypothesis of this project is that such an ANN can be designed to improve the performance of this instrument.

Minimizing errors in optical spectroscopic measurements: utilization of temperature control

The Optical Bridge/sup TM/ technique can be used to optically measure the concentration of theoretically any solute dissolved in an aqueous solution. This technique uses two individual fluids to measure the concentration of the solute in question. One fluid acts as a background, which contains very low concentrations (approx. zero) of the solute to be measured. The second fluid is identical to the first but contains higher concentrations of the solute in question. The difference in light absorbance between the two fluids is proportional to the concentration of the solute in question. However, like all optical measurements, the output from the Optical Bridge/sup TM/ is influenced by small temperature differences between these two fluids. Minimizing the influence that temperature has on the bridge output involved designing and constructing two separate temperature control systems. These systems were tested to determine which method kept the temperature difference between the two fluids within 0.1/spl deg/C. The achieved results indicate that, for temperature control, the water bath is more effective than adhesive heating tape.