NEXT-GENERATION MICRO-PHOTONIC TRANSDERMAL BIOSENSOR TECHNOLOGY FOR MAIN METABOLITE TRENDING AND MONITORING: IN-VIVO GLUCOSE DETECTION STUDY
- Augustinas Vizbaras, Lithuania
- Augustinas Vizbaras, Lithuania
- Ieva Šimonytė, Lithuania
- Arūnas Miasojedovas, Lithuania
- Tadas Bučiūnas, Lithuania
- Tomas Žukauskas, Lithuania
- Andreas De groote, Belgium
- Daan Martens, Belgium
- Žilvinas Dambrauskas, Lithuania
- Audrius Kučinskas, Lithuania
- Rimantas Stakauskas, Lithuania
- Vilma Zigmantaite, Lithuania
- Kristijonas Vizbaras, Lithuania
Abstract
Background and Aims
In this work we present initial validation results for novel optical biosensor technology based on gallium antimonide laser technology, which offers access to a largely unexplored 1.9-2.5 micron wavelength spectral region, containing molecule-specific ro-vibrational absorption bands and favorable skin transmission properties, allowing remote sensing of such molecules as glucose, lactate, urea, ethanol, etc. as has been demonstrated in previous proof-of-concept studies.
Methods
Here, we take a further step towards a real-life sensing scenario, and present experimental results of initial in vivo glucose detection study with pigs. Presented study contains data from more than 40 pigs. Each animal was sedated for the duration of the day and the blood glucose level was altered by means of intravenous glucose infusion. Optical data was gathered continuously by means of transdermal illumination and diffuse reflectance collection from the belly of the pig. In parallel, a blood sample was drawn and analyzed with a clinical blood analyzer, serving as a gold standard.
Results
To analyze the data, we have built a statistical linear regression-based algorithm involving machine learning for data grouping. Figure 1 illustrates excellent sensor performance of in vivo transdermal blood glucose level prediction in a wide dynamic concentration range for a single pig. Here, green dots represent data points that were used to build a PLS calibration model, whereas the red data points represent the validation data set.
Conclusions
We discuss the effects of the change of sensing position, motion, test subject, statistical batch size and provide outlook towards requirements for bringing the technology to market.