Wireless Wearable Sensor Paired With Machine Learning for the Quantification of Tissue Oxygenation

Abstract

The accurate knowledge of tissue oxygenation can be decisive for diagnostic applications in burns, limb injuries, and surgical interventions. Medical devices created for oxygenation measurements require extensive research and complex electronic, optical, and/or chemical techniques that typically result in nonmobile and expensive equipment. We have designed a wireless prototype that can detect changes in tissue oxygenation ( <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$pO_{2}$ </tex-math></inline-formula> ) making use of simple off-the-shelf electronic components and 3-D printing by measuring the phosphorescence intensity of an oxygen-sensing phosphor. The quantification of <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$pO_{2}$ </tex-math></inline-formula> was initially carried out by a phenomenological algorithm composed of a color-compensation matrix, a modified Stern–Volmer relation adding temperature dependence and an explicit photobleaching term. We improved the accuracy of measurement by employing a machine learning approach, which yields readings that are independent of changes in temperature and photobleaching, can be implemented into our data logging software, and potentially into the device’s firmware.