Biometric sensor optimization

     The discovery of the spectral absorption basis of CEi data occurred during bench testing of a prototype reflectance pulse oximeter for use on premature infants. Being unsure of the variables, the initial work explored the mechanical design of the sensor, including spacing of the apertures and prevention of light shunting from the emitters to the detector. The series of sensor modules also included a range of LED wavelengths, including 810 nm (isosbestic Infrared wavelength for hemoglobin-oxygen binding) and several "Red" wavelengths between 500 nm and 660 nm. We settled on 8 mm spacing of the apertures to produce an optimum combination of tissue interaction and detected signal strength, along with inclusion of a metal light barrier between the LEDs and the photodetector. When the resulting peak and trough raw data signals were analyzed using published methods, we found we had a remarkable signal strength of over 3.5% AC/Full Scale (commercial pulse oximeters typically operate well at 0.5 - 1.5%) and a credible SpO2 data output from our rather crude bench prototype. Further exploration of the raw data values during a hypoxic stress study resulted in the "first-light" waveform data graph also presented in this website.

     When development was resumed in 2014, we built a second-generation sensor system for further exploration. This version included four sensor channels, each using 660 nm Red LEDs, combined with one of a range of Infrared wavelength LEDs, from 810 nm to 850 nm to see which Infrared wavelength would produce the greatest signal response. With this USB cable-connected prototype, we were able to repeat the earlier hypoxia study, then explore responses to exercise and to increased oxygen breathing. Based on the remarkable results of those studies, we contracted to have a limited number of wearable versions built to overcome the limitations of the USB cable. Using this wearable version, we discovered several phenomena during sleep that were not detectable by pulse oximetry, and formalized these findings with a sleep laboratory study at UCSF and a published report.

     However, objective selection of the optimum LED wavelengths for detection of our new biometric remained in need of a scientifically credible investigation. As portrayed above, an optical fiber broadband light source (590 nm to 1000 nm) and optical fiber spectrometer system was utilized along with custom optical fiber cables and custom-designed skin interface components. A LabVIEW (National Instruments) control, display, and data recording program was created, as outlined above. The design intent of the study was to record the net spectral intensity variation during and following a brief period of ischemia of the subject's forearm. Based on this study, we have selected 685 +/- 10 nm (Red) and 850 +/- 10 nm (Infrared) as the optimum center wavelengths from which to derive the CEi data.

     From the above study, it is also apparent that 3 minutes of ischemia of the forearm changes the spectral absorbance of the skin, which becomes evident upon reperfusion with blood at the same SpO2 as before the ischemia. We interpret this net spectral absorption change as evidence of skin cellular adaptation to the extreme hypoxic stress induced by the period of ischemia. Specifically, both the magnitude and pattern of spectral absorption variation recorded in this study cannot be explained as due to changes in blood oxygen. The biomedical engineers on the FDA review panel during our online first pre-submission meeting were highly skeptical of our new information. However, our presentation of the combination of the published report of our UCSF sleep studies and this spectrometer study during our in-person second FDA pre-submission meeting resolved their skepticism, resulting in their formal acceptance of the science basis of our technology. They also noted that our technology, method of use, and biometric information is sufficiently different from pulse oximetry and NIRS that neither can be used as predicates. Because of this, the panel requested that we provide a De Novo submission for our Cellular Energy Monitor technology as a new Class II medical device.