Devices, systems, and methods for unbiased pulse oximetry

Abstract

Disclosed herein are devices, systems, and methods for reducing skin pigmentation bias in pulse oximetry measurements. The systems and methods can include devices, systems, and methods for narrowing the full-width-half-maximum bandwidth of light used to calculate SpO2.

Description

PATENT Atty Docket: 093698-822591 Via Patent Center DEVICES, SYSTEMS, AND METHODS FOR UNBIASED PULSE OXIMETRY CROSS-REFERENCE TO RELATED APPLICATIONS

[0001] This application claims the benefit of U.S. Application No. 63 / 596,776, filed November 7, 2023, the entire contents of which are incorporated herein by reference in their entirety. FIELD

[0002] The present disclosure relates to devices, systems, and methods for reducing or preventing bias in pulse oximetry measurements. GOVERNMENT SUPPORT

[0003] This invention was made with government support under 1 R01 EB033799- 01 awarded by the National Institute of Biomedical Imaging and Bioengineering. The government has certain rights in the invention. BACKGROUND

[0004] Empirical studies over the past 15 years have shown that measurements of arterial oxygen saturation (SaO2) by pulse oximeters are affected by skin pigmentation despite there being no widely known or published explanation for why this occurs. To the contrary, literature on the operation of pulse oximeters often state that such measurements are not affected by skin pigmentation. While skin pigmentation bias is present in many pulse oximeters, there have been few, if any, studies conducted on why skin pigmentation bias is present or how to solve this problem.

[0005] Hypoxemia (low levels of oxygen in blood) can be treated with quick and accurate testing. However, failures of current market pulse oximeters have led to a failure to diagnose hypoxemia due to erroneously over-estimating SpO2 measurements. Further, the failures of pulse oximeters have been exasperated by COVID-19.

[0006] Therefore, there is a need for devices, systems, and methods to reduce or prevent skin pigmentation bias in pulse oximeters. SUMMARY OF INVENTION Page 1 of 38 99391552.2PATENT Atty Docket: 093698-822591 Via Patent Center

[0007] The present application relates to a device for measuring oxygen saturation in blood. The device can include a first light source, a second light source, a photodetector, and filters. In an aspect, the first light source can be configured to emit a first light having a wavelength of about 630 to about 700 nm through a body part of a patient. In an aspect, the second light source can be configured to emit a second light having a wavelength of about 820 nm to about 1000 nm through the body part of the patient. In an aspect, the photodetector can be located on an opposite side of the body part from the first light source and the second light source, the photodetector configured to receive the first light and the second light. In an aspect, the filter can be configured to narrow a full-width-half-maximum (FWHM) bandwidth of the first light and / or the second light a narrow FWHM bandwidth. In an aspect, the filter can be a physical filter, an inherent filter to the first and / or second light source, or an electronic filter in communication with a spectroradiometer and a processor. With regard to this last approach, a white light source can be used instead of the first and second light sources described above. In another aspect, the first light source and the second light source are rapidly counter phased.

[0008] In various aspects, the filter narrows the FWHM bandwidth of the first light and / or the second light before being transmitted through the body part. In another aspect, the filter narrows the FWHM bandwidth of the first light and / or the second light received at the photodetector after being transmitted through the body part. In an aspect, the processor is configured to receive data from the spectroradiometer, filter the data, via the electronic filter, to narrow the FWHM bandwidth, calculate an oxygen saturation value, and output the oxygen saturation value on a display in communication with the processor.

[0009] In various aspects, the first light source and / or second light source is a laser diode or resonant-cavity LED and the inherent filter is a resonant cavity in the laser or resonant-cavity LED. In an aspect, the physical filter is a narrowband interference filter. In another aspect, the narrow FWHM bandwidth is about 10 nm, about 5 nm, or less. In another aspect, the first light source and the second light source comprise an LED, a laser diode, or combinations thereof. In another aspect, the device removes skin pigmentation bias from oxygen saturation measurements.

[0010] Further provided herein is a method for measuring oxygen saturation in blood. The method can include providing, via a first light source, a first light having a Page 2 of 38 99391552.2PATENT Atty Docket: 093698-822591 Via Patent Center wavelength of about 630 nm to about 700 nm to a body part of a patient, providing, via a second light source, a second light having a wavelength of about 820 nm to about 1000 nm to the body part of the patient, receiving the first light and the second light at a photodetector located adjacent to the first light source and second light source or located on an opposite side of the body part from the first light source and the second light source, filtering the first light and / or the second light to a narrow FWHM bandwidth, and measuring the oxygen saturation in blood of the patient based on the first light and the second light received at the photodetector. In an aspect, the first light source and the second light source are rapidly counter phased.

[0011] In various aspects, filtering the first light and / or the second light comprises using a filter before the first light and the second light are transmitted through the body part of the patient. In another aspect, filtering the first light and / or the second light comprises using a filter after the first light and the second light are transmitted through the body part of the patient. In an aspect, the filter is a narrowband interference filter, an inherent filter to the first and / or second light source or using a white light source and an electronic filter in communication with a spectroradiometer and a processor. In another aspect, the first light source and / or second light source is a laser diode or resonant-cavity LED and the inherent filter is a resonant cavity in the laser or resonant- cavity LED configured to filter the first light and / or second light. In an aspect, the narrow FWHM bandwidth is about 10 nm, about 5 nm, or less. In an aspect, filtering the first light and / or second light comprises using a physical filter directly deposited on the first light source and / or second light source or the physical filter is an encapsulate material configured to narrow the FWHM bandwidth. In an aspect, the FWHM bandwidth removes skin pigmentation bias from oxygen saturation measurements.

[0012] Further provided herein is a device for measuring oxygen saturation in blood. The device can include a first light source configured to emit a first light having a wavelength of about 630 nm to about 700 nm and a FWHM bandwidth of about 5 nm through a body part of a patient, a second light source configured to emit a second light having a wavelength of about 820 nm to about 1000 nm through a body part of a patient, and a photodetector located either adjacent to the first light source and the second light source in a reflectance mode or located on an opposite side of the body part from the first light source and the second light source in a transmittance mode, the photodetector configured to receive the first light and the second light. In an aspect, Page 3 of 38 99391552.2PATENT Atty Docket: 093698-822591 Via Patent Center the first light source is a resonant cavity LED, an LED having an interference filter, or a laser diode.

[0013] Further provided herein is a device for measuring oxygen saturation in blood. The device can include a first light source configured to emit a first light having a wavelength of about 630 nm to about 700 nm through a body part of a patient, a second light source configured to emit a second light having a wavelength of about 820 nm to about 1000 nm through a body part of a patient, a photodetector located adjacent the first light source and the second light source or located on an opposite side of the body part from the first light source and the second light source, the photodetector configured to receive the first light and the second light, and a filter attached to the photosensor, the filter configured to narrow at least the first light received by the photosensor to a FWHM bandwidth of about 5 nm or less. In an aspect, the first light source and the second light source are rapidly counter phased. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] For a more complete understanding of this disclosure and its features, reference is now made to the following description, taken in conjunction with accompanying drawings, in which like reference numerals denote like elements:

[0015] FIG.1 is a graph showing probability of patients to be discharged early with SpO2 levels greater than 90 and less than or equal to 90, where the x-axis is days since hospital arrival.

[0016] FIG. 2 is a diagram showing filtering of LEDs having peak emission wavelengths of 660 nm and 900 nm in an exemplary device.

[0017] FIG.3 is a graph of transmittance percentage and wavelength.

[0018] FIG.4 is a graph of wavelength and temperature of a resonant cavity LED.

[0019] FIG.5 is a graph of wavelength and temperature of a laser diode.

[0020] FIG.6 is a graph of an exemplary system in one example.

[0021] FIG.7 is a graph the correlation between the bivariate slope of SpO2 readings (wide v. narrow bandwidth) and the ratio of finger transmittance at 660 nm and 900 nm. The closed symbols in the unshaded portion of the graph have transmittance ratios greater than the median value (1.1) and those in the shaded portions have ratios Page 4 of 38 99391552.2PATENT Atty Docket: 093698-822591 Via Patent Center less than the median. The open symbols represent data collected from a single subject on two different days.

[0022] FIG. 8 is a graph of arterial oxygen saturation (%) v. oxygen saturation on pulse oximetry (%) for groups of white patients and groups of black patients.

[0023] FIG.9 is a flow chart illustrating skin pigmentation bias.

[0024] FIG. 10 is a graph of absorption coefficient in relation to wavelength for Hb and HbO2.

[0025] FIG. 11 is a graph of wavelengths of light emitted by a polychromatic light source and a FWHM bandwidth narrowed light source.

[0026] FIG.12 is a graph of SpO2 and SaO2 for a modeled light source with a FWHM bandwidth filter and a light source without a FWHM bandwidth filter.

[0027] FIG.13 is an experimental setup to determine SpO2 bias and the effect of a FWHM bandwidth filter in one example.

[0028] FIG.14 is a chart of an experimental setup to determine SpO2 bias and the effect of a FWHM bandwidth filter in one example.

[0029] FIG.15A is a graph of experimental results of SpO2 readings using a FWHM bandwidth filtered light source and an unfiltered polychromatic light source for a highly pigmented subject.

[0030] FIG.15B is a graph of the SpO2 readings using a FWHM bandwidth filtered light source and an unfiltered polychromatic light source for a highly pigmented subject.

[0031] FIG.15C is a graph of experimental results of SpO2 readings using a FWHM bandwidth filtered light source and an unfiltered polychromatic light source for a lightly pigmented subject.

[0032] FIG.15D is a graph of the SpO2 readings using a FWHM bandwidth filtered light source and an unfiltered polychromatic light source for a lightly pigmented subject.

[0033] FIG. 16 is a graph of arterial oxygen saturation (%) v. oxygen saturation on pulse oximetry (%) for groups of white patients and groups of black patients showing skin pigmentation bias of pulse oximeters for patients with low melanin contents (white patients) and high melanin contents (black patients). Page 5 of 38 99391552.2PATENT Atty Docket: 093698-822591 Via Patent Center

[0034] FIG.17 is a graph of spectra output or transmittance, absorption coefficient, and wavelength illustrating melanin absorptance, melanin transmittance, and changing absorption over output range changes the effective emission wavelength.

[0035] FIG.18 is a zoomed in view of FIG.17.

[0036] FIG.19 is a graph of modulation ratio (R) used to calculate SpO2 in relation to SaO2.

[0037] FIG.20 is a graph of modeled SpO2 v. actual SaO2 for a FWHM filtered light source and an unfiltered polychromatic light source.

[0038] FIG.21 is a chart of percentage error in SpO2 readings for selected FWHM bandwidths of light sources using a prediction model.

[0039] FIG. 22 is a diagram of the geometry used in modeling optical signals and detection for SpO2 readings.

[0040] FIG. 23 illustrates equations for modeling SpO2 readings from pulse oximeters.

[0041] FIG.24 illustrates equations for calculating SaO2.

[0042] FIG. 25 is a graph showing SaO2, light attenuation, and time for red light intensity and infrared light intensity.

[0043] FIG. 26 is a graph of normalized spectral power and wavelength of a pulse oximeter having a FWHM bandwidth of 22 nm.

[0044] FIG. 27 is a graph of normalized spectral power and wavelength of a pulse oximeter having bandwidth of 16 nm and a pulse oximeter having a filtered FWHM bandwidth of 10 nm.

[0045] FIG. 28 is a graph the correlation between the bivariate slope of SpO2 readings (wide v. narrow bandwidth) and the ratio of finger transmittance at 660 nm and 900 nm. The closed symbols in the unshaded portion of the graph have transmittance ratios greater than the median value (1.1) and those in the shaded portions have ratios less than the median. The open symbols represent data collected from a single subject on two different days.

[0046] FIG. 29 is an overhead view of an experimental setup, showing a subject wearing a breathing mased connected to a Reduced Oxygen Breathing Device (ROBD2) with SpO2 and spectrally resolved photoplethysmography (srPPG) finger probes. Page 6 of 38 99391552.2PATENT Atty Docket: 093698-822591 Via Patent Center

[0047] FIG. 30 illustrates spectral power distributions of polychromatic (high bandwidth) and nearly monochromatic (low bandwidth) virtual sources used for bandwidth sensitivity calculations.

[0048] FIG. 31 illustrates R values for polychromatic (ordinate) and narrow band (abscissa) spectra from one subject. The slope of the least squares regression (0.9414) is the R-slope metric for this subject.

[0049] FIG. 32 is a graph of R-slope v. melanin fraction. A higher melanin fraction (abscissa) indicates more heavily pigmented skin, and a lower R-slope (ordinate) indicates greater sensitivity to spectral bandwidth of the light sources. A significant negative correlation (P=0.00008) indicates that higher melanin concentrations are more sensitive to spectral bandwidth.

[0050] FIG. 33 illustrates whisker plots for the first and fourth quartiles of subjects ranked by melanin fraction (the most lightly pigmented compared with the most heavily pigmented subjects), showing that more heavily pigmented subjects are more sensitive to spectral bandwidth. Two-tailed Welch’s t-test, P=0.0011.

[0051] FIG. 34 is a graph illustrating measurement-to-measurement uncertainty of the R-slope value (ŠR-slope) vs. melanin fraction for each subject. DETAILED DESCRIPTION

[0052] It will be appreciated that for simplicity and clarity of illustration, where appropriate, reference numerals have been repeated among the different figures to indicate corresponding or analogous elements. In addition, numerous specific details are set forth in order to provide a thorough understanding of the examples described herein. However, it will be understood by those of ordinary skill in the art that the examples described herein can be practiced without these specific details. In other instances, methods, procedures, and components have not been described in detail so as not to obscure the related relevant feature being described. Also, the description is not to be considered as limiting the scope of the embodiments described herein. The drawings are not necessarily to scale and the proportions of certain parts may be exaggerated to better illustrate details and features of the present disclosure.

[0053] Several definitions that apply throughout the above disclosure will now be presented. The term “coupled” is defined as connected, whether directly or indirectly Page 7 of 38 99391552.2PATENT Atty Docket: 093698-822591 Via Patent Center through intervening components, and is not necessarily limited to physical connections. The connection can be such that the objects are permanently connected or releasably connected. The term “substantially” is defined to be essentially conforming to the particular dimension, shape or other word that substantially modifies, such that the component need not be exact. The terms “comprising,” “including” and “having” are used interchangeably in this disclosure. The terms “comprising,” “including” and “having” mean to include, but not necessarily be limited to the things so described.

[0054] The term “filter” is defined as physical filters (e.g., interference filters), inherent filters (e.g., laser diodes and resonant cavity LEDs inherently have an interference filter), and electronic filters that filter data received by a spectroradiometer and a processor. Electronic filters can be filters that work within a processor and filter light received at a photodetector to a narrow bandwidth.

[0055] Disclosed herein are devices, systems, and methods for reducing or preventing skin pigmentation bias in pulse oximeters. Skin pigmentation bias occurs due to polychromatic light sources being used in pulse oximeters. The spectral power distribution of these polychromatic light sources are transformed due to the spectral absorption of melanin, the magnitude of the transformation depends upon melanin concentration (e.g., skin tones). Therefore, polychromatic light sources naturally bias pulse oximetry results, leading to less accurate SpO2 measurements in patients depending upon the melanin concentrations in their skin. The present systems and methods remove or substantially remove skin pigmentation bias in SpO2 measurements.

[0056] The result of skin pigmentation bias in pulse oximeters leads to inaccurate measurements of SpO2. When SpO2 is inaccurate, diagnosis and treatment of a variety of health issues can be inadequate. As illustrated in FIG. 1, SpO2 measurements, which in turn lead to hypoxia are vital to health outcomes. Individuals with SpO2 measurements above 90 are more likely to be discharged from a hospital alive than individuals with SpO2 measurements below 90. The SpO2 readings directly inform clinical decisions that affect patient medical outcomes. For example, pulse oximeters can be used at a hospital, on the way to the hospital, or at home. If the SpO2 measurements are inaccurate, the patient may not receive adequate care. For example, the patient may not receive oxygen treatments in time if the SpO2 reading is Page 8 of 38 99391552.2PATENT Atty Docket: 093698-822591 Via Patent Center inaccurate (e.g., if the SpO2 reading erroneously measures a higher SpO2 than an actual SaO2 value).

[0057] Despite the findings of several studies, more recent empirical have documented that commercially available pulse oximeter measurements (SpO2) systematically overestimate arterial oxygen saturation levels (SaO2) for persons with dark skin at reduced oxygen saturation levels. This is where the accuracy of pulse oximetry is most critical. Because physicians heavily rely on pulse oximeter readings for diagnosis and treatment, this systematic bias practically means that individuals with relatively high concentrations of melanin in the skin might not reach the threshold for hospitalization, escalation of care, or corrective interventions (e.g., where SpO2 is 88- 92%), thereby putting this group at greater risk for occult hypoxemia and higher mortality than those with lightly pigmented skin. Although the problem has now become more widely recognized, there has been no satisfactory published explanation for this systematic bias.

[0058] Recent studies have reinforced the growing concern with biased pulse oximeter (SpO2) readings from dark-skin patients with low levels of arterial oxygen saturation (SaO2). However, these studies do not address the potential source of the systemic bias in pulse oximetry. The systematic bias comes from the polychromatic light-emitting diode (LED) light sources utilized in all commercially available pulse oximeters.

[0059] Two light sources and a photodiode are used in most pulse oximetry systems. The two light sources, one to generate red light and one to generate infrared light, are alternately energized and the photodiode measures and records how much light is transmitted, typically through a finger, throughout the pulse cycle. Two equations are used to measure SpO2. Equation 1 is the ratio of the ratio of transmitted red light and the ratio of transmitted infrared light. Equation 2 calculates SpO2. ^^^^^^^^^^ ^^ =^^^^^^^^^^ ^^^^^^^^ (Eq.1) ^^^^^^^^ SpO2 = 110 – 25R (Eq.2) Page 9 of 38 99391552.2PATENT Atty Docket: 093698-822591 Via Patent Center

[0060] R is the ratio of two ratios, one for the transmitted red (red) light and one for the transmitted infrared (ir) light during the complete pulse cycle. Peak amplitudes occur between heart beats when there is less blood and minimum amplitudes occur with each heartbeat as light-absorbing blood expands the capillaries. The pulse cycle waveform therefore has a time-varying component (alternating current, AC) superimposed on its average value (direct current, DC). For calibration of a specific type of pulse oximeter, a population cohort is selected, and their blood oxygen levels are varied with a controlled breathing apparatus. Values of R are determined at different blood oxygen levels, measured directly by arterial blood gas analysis, which are then used to calibrate the SpO2 values generated by that type of pulse oximeter.

[0061] In other aspects, R can be calculated using Equation 3. ^^ =log (^^^^^^^^^^) log (^^^^^^^^) (Eq.3)

[0062] In some aspects, the equation to calculate SpO2 can be higher-order equations for calculating SpO2. The equation to calculate SpO2 can be any equation known in the art or used in commercial pulse oximeters. Even more complex SpO2 calculations result in skin pigmentation bias.

[0063] Studies have consistently stated that pulse oximeters utilize monochromatic wavelengths. However, current market pulse oximeters utilize polychromatic wavelengths. Commonly, the spectral linewidth of LEDs, measured over the range of wavelengths where the emission intensity is greater than half of its peak, is on the order of 25 nm for red and typically greater for near infrared emitters. Therein lies the source of systematic measurement bias. Spectral absorption of melanin interacts with the polychromatic LED light sources, resulting in a positive drift from the calibration curve at higher melanin concentrations.

[0064] Pulse oximeters are calibrated using a cohort representative of the US population. Therefore, by definition, minorities are not equally represented in the calibration process. For polychromatic sources such as LEDs, the presence of melanin in relatively high concentrations will effectively redshift the center wavelength of the transmitted light. This redshift throws off the calibration based on light-skin cohorts, resulting in positive measurement bias for dark-skin individuals. The measurement bias translates into systematically higher than actual estimates of O2 saturation of hemoglobin. The higher the concentration of melanin in the skin, the larger the Page 10 of 38 99391552.2PATENT Atty Docket: 093698-822591 Via Patent Center magnitude of the redshift, leading to larger overestimation errors for individuals with darker skin.

[0065] If pulse oximeter calibrations had been done with a predominantly dark-skin cohort, the reverse bias would occur. Those with light skin would show systematically lower estimates of O2 saturation of hemoglobin. The problem, then, cannot be solved simply by asking the FDA to change sampling requirements for pulse oximeter calibration. Any change in the sampling requirements for a given cohort will disadvantage some other segment of the population. And placing individuals seeking medical treatment in the correct group is nearly impossible because the appearance of skin reflectance does not necessarily reflect actual melanin absorption in the far red and infrared spectral regions. The use of monochromatic, or nearly monochromatic, light sources will eliminate the bias problem, positive or negative, associated with current pulse oximeters.

[0066] The devices and systems disclosed herein can include a first light source, a second light source, a photodetector, and one or more filters or, alternatively, a white light source, spectroradiometer and processor. The first and second light sources can emit a light that enters the body part of the patient and reflects back towards the photodetector located near (e.g., adjacent to) the first and second light sources. In another example, the first and second light sources can emit a light that transmits through the body part and is measured by the photodetector on the opposite side of the body part from the first and second light sources. In an example, the first light source can emit a first light having a wavelength of about 630 nm to about 700 nm. In some examples, the first light source can emit a first light having a wavelength of about 630 nm to about 650 nm, about 650 nm to about 670 nm, or about 670 nm to about 700 nm. In an example, the first light source can emit a first light have a wavelength of about 660 nm. In an example, the second light source can emit a second light can having a wavelength of about 820 nm to about 1000 nm. In another example, the second light can have a wavelength of about 820 nm to about 840 nm, about 840 nm to about 860 nm, about 860 nm to about 890 nm, or about 890 nm to about 910 nm. In another example, the second light can have a wavelength of about 900 nm. In some examples, the first light source and the second light source can be rapidly counter phased, meaning the first light and the second light are emitted at different times. Page 11 of 38 99391552.2PATENT Atty Docket: 093698-822591 Via Patent Center

[0067] The photodetector can be operable to receive the first light and the second light. In some examples, the photodetector can be located adjacent to the first light source and the second light source to receive reflected light from the body part. In other examples, the photodetector can be located on an opposite side of the body part from the first light source and the second light source to receive the first light and the second light transmitted through the body part.

[0068] In an aspect, when a white light source is used with a spectroradiometer and a processor, white light needs to be decomposed into individual wavelengths and then processed. The white light source can be used as an alternative to the first light source and the second light source while operating to transmit light into the body part.

[0069] The system can further include a processor configured to receive data from a spectroradiometer and filter the data with the electronic filter to narrow the FWHM (full- width-half-maximum) bandwidth of the white light emitted by the white light source, calculate an oxygen saturation value, and output the oxygen saturation value on a display in communication with the processor. In another example, when the first light source and the second light source are used, the processor can be configured to receive information from the photodetector, calculate an oxygen saturation value, and output the oxygen saturation value on a display in communication with the processor.

[0070] The filter can be configured to narrow a FWHM bandwidth of the first light and / or the second light to a narrow FWHM bandwidth. As described herein, the narrow FWHM bandwidth can reduce or remove skin pigmentation bias SpO2 readings. The narrow FWHM bandwidth can be about 1 nm to about 10 nm. For example, the narrow FWHM bandwidth can be about 1 nm to about 5 nm, about 5 nm to about 10 nm. The filter can be a physical filter such as an interference filter. In an example, the interference filter can be a dielectric filter. An interference filter can be placed directly in front of the first light source and / or second light source such that the first light and / or second light are filtered prior to being transmitted to the body part. In another example, the interference filter can be located proximal the photodetector such that the first light and / or second light is filtered after being transmitted through the body part or reflected from the body part. In some examples, both the first light source and the second light source can have an interference filter located directly in front of the first light source and the second light source. In other examples, the filter can be a dual-pass filter. The dual-pass filter can both narrow the bandwidth of one source and still transmit light Page 12 of 38 99391552.2PATENT Atty Docket: 093698-822591 Via Patent Center from the other. The dual-pass filter can be used with the first light source and the second light source.

[0071] FIG. 2 illustrates the use of interference filters on the first light source. The interference filter is placed perpendicular to the optical path of the first light source. The first light can have a peak emission wavelength of about 660 nm and the second light can have a peak emission wavelength of about 900 nm. The transmittance of the first light is greater than 0.8 (80%). The interference filter can narrow the FWHM bandwidth of the first light to about 8 nm (e.g., wavelengths between about 656 nm to about 664 nm). The transmittance of the second light is greater than 0.5 (50%).

[0072] FIG. 3 illustrates the transmittance of the first light source using an interference filter located proximal the first light source and perpendicular to the optical path of the first light source. The transmittance of the first light after being filtered by the interference filter was within a narrow FWHM bandwidth of about 5 nm.

[0073] In another example, the filter can be inherent in the first light source and / or the second light source. For example, the first light source and / or second light source can be a laser diode. The laser diode inherently has a filter in its resonant cavity, thereby providing a narrow FWHM bandwidth. In another example, the first light source and / or the second light source can be resonant-cavity LEDs which inherently has a filter in the resonant cavity configured to emitter a narrow FWHM bandwidth.

[0074] In another aspect, the filter can be a physical filter on the first light source and / or the second light source. The physical filter can be directly deposited on the first light source and / or second light source configured to narrow the FWHM bandwidth. In another example, the physical filter can be an encapsulate material on the first light source and / or the second light source configured to narrow the FWHM bandwidth.

[0075] FIG.4 illustrates the dominant bandwidth changes for a resonant cavity LED based on temperature. The dominant bandwidth only changes + / - 2 nm in normal operating temperatures (e.g., room temperature). Therefore, a resonant cavity LED can be used instead of an interference filter to produce the same narrow FWHM bandwidth necessary to reduce or prevent skin pigmentation bias.

[0076] FIG. 5 illustrates wavelength shift of laser diodes as a function of case temperature. The wavelength shift per temperature is on the scale of picometers, and therefore has no effect on the operation of laser diodes as a light source providing a narrow FWHM bandwidth. Page 13 of 38 99391552.2PATENT Atty Docket: 093698-822591 Via Patent Center

[0077] In another example, the filter can be an electronic filter. The electronic filter can be in communication with a spectroradiometer and a processor. The electronic filter can be configured to extract data at a narrow FWHM bandwidth. This data can then be processed for the narrow FWHM bandwidth by the processor to output accurate SpO2 readings.

[0078] FIG. 6 illustrates an exemplary system for pulse oximetry. The exemplary system can include a continuous broadband source, a fiberoptic cable, a grating spectrometer, a CCD array detector, and a processor. In some examples, the system can include a white light source or white light sources and a spectroradiometer. In this example, the continuous broadband source can be a white light source. The white light source can be located on one side of a body part of a patient (e.g., finger). The fiberoptic cable can be located on the opposite side of the body part and be configured to receive the light emitted by the with light source. In another example, the fiberoptic cable can be located on the same side of the body part as the white light source in a reflectance mode (e.g., the fiberoptic cable receives reflected light from the body part). The grating spectrometer can be used to extract the desired wavelengths of light. For example, the grating spectrometer can extract a first light having a wavelength of about 660 nm and a second light having a wavelength of about 900 nm. The light data can then be processed by a CCD detector to produce images. The images can then be processed by the processor to determine the SpO2 measurement for the patient. This system can provide accurate SpO2 measurements by significantly narrowing the FWHM bandwidth of the light source (e.g., by extracting the relevant wavelengths using the grating spectrometer).

[0079] In an aspect, the first light source should provide 90% of the first light optical power within 5 nm of the center wavelength of the first light (e.g., 660 nm) and 99.9% of the first light optical power within 15 nm of the center wavelength of the first light (e.g., 660 nm). The second light source should provide 90% of the second light source optical power within 30 nm of the center wavelength of the second light (e.g., 900 nm) and 99.9% of the second light optical power within 60 nm of the center wavelength of the second light (e.g., 900 nm). In an example, all spectral power measurements are obtained with a spectroradiometer with a spectral resolution no greater than 5 nm and a spectral measurement uncertainty within 1 nm. In an example, the minimum optical Page 14 of 38 99391552.2PATENT Atty Docket: 093698-822591 Via Patent Center power generated by the first light source and the second light source should be at least 1 mW.

[0080] Further provided herein is a method for measuring an oxygen saturation in blood. The method can include providing, via a first light source, a first light having a wavelength of about 630 nm to about 700 nm to a body part of a patient, providing, via a second light source, a second light having a wavelength of about 820 nm to about 1000 nm to the body part of the patient, receiving the first light and the second light at a photodetector located adjacent to the first light source and second light source or located on an opposite side of the body part from the first light source and the second light source, filtering the first light and / or the second light to a narrow FWHM bandwidth, and measuring the oxygen saturation in blood of the patient based on the first light and the second light received at the photodetector. The first light source and the second light source can be counter phased to use a single photodetector with a rapid response time. The method can utilize any of the devices, systems, or components described herein.

[0081] In various aspects, filtering the first light and / or the second light comprises using a filter before the first light and the second light are transmitted through the body part of the patient. In another aspect, filtering the first light and / or the second light comprises using a filter after the first light and the second light are transmitted through the body part of the patient. In an aspect, the filter is a narrowband interference filter, an inherent filter to the first and / or second light source, or an electronic filter in communication with a processor. In another aspect, the first light source and / or second light source is a laser diode or resonant-cavity LED and the inherent filter is a resonant cavity in the laser or resonant-cavity LED configured to filter the first light and / or second light. In an aspect, the narrow FWHM bandwidth is about 10 nm, about 5 nm, or less. In an aspect, filtering the first light and / or second light comprises using a physical filter directly deposited on the first light source and / or second light source or the physical filter is an encapsulate material configured to narrow the FWHM bandwidth. In an aspect, the FWHM bandwidth removes skin pigmentation bias from oxygen saturation measurements. EXAMPLES Example 1: Page 15 of 38 99391552.2PATENT Atty Docket: 093698-822591 Via Patent Center

[0082] To validate theoretical analysis that spectral absorption of melanin interacts with polychromatic LED sources, resulting in a positive drift from the calibration curve at higher melanin concentrations, a study was conducted that delivered controlled respiratory oxygen concentrations to individuals of different skin pigmentations. Simultaneous R value readings were obtained from two finger probes, one from a commercially available probe (wideband) and the other from a similar device modified with a narrow-band spectral filter (12 nm full width at half maximum of 660 nm, unaltered LED bandwidth pass of ~60 nm centered at 900 nm). Simultaneous R value readings were obtained from the two probes worn on the fifth distal phalanx of the left and right hands while the subject breathed a range of well-controlled O2 concentrations (21-9.5%) over a 20-min protocol. Individual slopes for each of the 10 subjects were determined by plotting the R value readings from the wideband probe against the R value readings from the narrowband probe over the entire range of O2 concentrations. If, as hypothesized, the narrowband probe reduced the systematic bias as a result of skin pigmentation, the bivariate slope values for individuals with dark skin should be systematically shallower than those for people with light skin.

[0083] Absolute finger transmittance values at 660 nm and at 900 nm were measured and the ratio of these two values was determined for each subject. Individuals with dark skin should have lower ratios because melanin, the primary pigment that affects skin color, absorbs more of the 660 nm light without appreciably affecting 900 nm light absorption. The median 660 nm / 900 nm ratio was 1.1, arbitrarily separating dark-skin from light-skin subjects for the analysis. FIG. 7 shows that the bivariate slope value increases as skin lightness increases. The mean slope of the lighter skin subjects was greater than that for the darker skin subjects (one-tail Student t-test: P=0.009). It is important to note that this empirical approach avoids having to measure absolute O2 partial pressure in arterial blood. Only the relative bias, that is, the change in sensitivity (slope) as a result of skin pigmentation, was needed to validate the theoretical analysis. This study shows that the skin pigmentation bias in current market pulse oximeters can be virtually eliminated when narrowband light sources are used. Example 2:

[0084] Hypoxemia is a condition where a patient has a low level of oxygen in their blood. Hypoxemia is generally associated with poorer health outcomes, which was further amplified by COVID-19. Pulse oximetry is universally utilized in hospital, pre- Page 16 of 38 99391552.2PATENT Atty Docket: 093698-822591 Via Patent Center hospital, and home settings for non-invasive monitoring of oxygen levels in blood (SO2). These SpO2 readings directly inform critical decisions that affect patient outcomes. For example, SpO2 readings can directly affect hospital admission and ICU transfer as well as application of and home prescription for supplemental oxygen. Unrecognized and untreated hypoxemia leads to poor care and bad outcomes. FIG.1 illustrates cumulative incidence functions for the probability of being discharged alive by out-of-hospital SpO2 level. As illustrated, patients with higher SpO2 levels have a much greater likelihood of being discharged from a hospital alive, showing the importance of accurate SpO2 readings.

[0085] FIG. 8 illustrates SaO2 as a function of SpO2 for white patients and black patients. As illustrated, SpO2 readings tend to significantly overestimate actual blood oxygen levels in black patients as compared to white patients. Pulse oximetry readings are biased higher for people with darkly pigmented skin. Hypoxemia is not flagged by SpO2 at a much higher rate for black patients (11.4%) than white patients (3.6%).

[0086] Past modeling efforts for pulse oximeters assumed monochromatic red light and infrared light sources. However, LEDs in pulse oximeters are polychromatic. FIG. 9 illustrates an example flow chart of the skin pigmentation bias problem in simple terms, polychromatic light combines with melanin spectral absorption to produce skin pigmentation bias in current market pulse oximeters. FIG.10 illustrates pulse oximetry optical signals. As illustrated, the LED sources are not monochromatic as the output range is greater than 60 nm. The absorption coefficients of Hb and HbO2 are also illustrated. As show, the absorption coefficients have steep slopes and the polychromatic nature of the LED sources causes different absorption coefficients throughout the output range, leading to inaccurate results.

[0087] A solution to the skin pigmentation bias problem includes narrowing LED spectral linewidth (e.g., FWHM bandwidth) to eliminate dependence on spectral absorption. In this manner skin pigmentation bias is eliminated and calibration accuracy is improved. FIG.11 illustrates an LED with a narrow spectral linewidth 1100 in comparison to current polychromatic light sources 1102. As illustrated, the narrow spectral linewidth 1100 only has a linewidth of about 15 nm, whereas the polychromatic light source 1102 has a linewidth of about 80 nm to about 100 nm. FIG.12 illustrates the difference in SpO2 and SaO2 readings using light sources with a narrow linewidth 1200 (FWHM bandwidth) and a commercially available polychromatic light source Page 17 of 38 99391552.2PATENT Atty Docket: 093698-822591 Via Patent Center 1202 using an SpO2 prediction model. The narrow linewidth was accomplished through the use of a linewidth narrowing optical filter. As illustrated, the commercially available polychromatic light sources had significantly higher SpO2 readings than actual SaO2 readings. The bias was significantly higher at lower SaO2 values, meaning that commercially available polychromatic sources overestimate SaO2, thereby misdiagnosing hypoxemia and other blood oxygen related conditions.

[0088] An experiment was conducted to determine the effects of a narrow linewidth LED instead of commercial polychromatic LEDs used in commercial pulse oximeters. The experiment included two pulse oximeters. One pulse oximeter was a commercial pulse oximeter having a FWHM bandwidth (i.e., linewidth) of 16 nm (i.e., wide LED linewidth pulse oximeter). The other pulse oximeter was fitted with a filter narrowing the linewidth to a 10 nm FWHM bandwidth (i.e., narrow LED linewidth pulse oximeter). FIG. 13 illustrates the experimental set up. The subject was provided low O2 air (simulated altitude) and SpO2 measurements were taken with both devices. FIG.14 illustrates the two subject groups, both containing over 30 subjects. Subject group 1 consisted of subjects with lightly pigmented skin. Subject group 2 consisted of subjects with highly pigmented skin. In other examples, other experimental setups similar to the setup shown in FIG.13 can be used.

[0089] FIG. 15A illustrates SpO2 measurements over time for both the wide LED linewidth pulse oximeter and the narrow LED linewidth pulse oximeter for a highly pigmented subject. FIG.15B illustrates a comparison of SpO2 readings between the wide LED linewidth pulse oximeter and the narrow LED pulse oximeter for a highly pigmented subject. FIG. 15C illustrates SpO2 measurements over time for both the wide LED linewidth pulse oximeter and the narrow LED linewidth pulse oximeter for a lightly pigmented subject. FIG.15D illustrates a comparison of SpO2 readings between the wide LED linewidth pulse oximeter and the narrow LED pulse oximeter for a highly pigmented subject. As illustrated by FIG. 15A, the highly pigmented subject had a consistently lower SpO2 reading when breathing high O2 air and then a consistently lower SpO2 reading when breathing low O2 air as measured by the wide LED linewidth pulse oximeter compared to the narrow LED pulse oximeter. As illustrated in FIG.15C the lightly pigmented subject had relatively similar SpO2 readings between both the wide LED linewidth pulse oximeter and the narrow LED linewidth pulse oximeter. The Page 18 of 38 99391552.2PATENT Atty Docket: 093698-822591 Via Patent Center difference between the dark and light skin subjects is reflected in the difference in slopes relating narrow band to wide band sources as illustrated in FIGS.15B and 15D. Example 3:

[0090] FIG. 16 illustrates SaO2 as a function of SpO2 for white patients and black patients using a commercial pulse oximeter having an LED bandwidth of 19 nm and a peak wavelength of 660 nm for red light emission. The paired measurements correspond to the boxes on the graph. For example, the first white patient box had 92 white patients and the first black patient box had 20 patients, and so on. FIG.16 clearly illustrates the bias in pulse oximeter readings.

[0091] FIGS. 17-18 illustrate how melanin absorption affects LED emission. As illustrated, as the wavelength of light emitted increases, the melanin absorption decreases. While these relationships are factored into the current operating calibration models, these models are naturally biased towards lightly pigmented individuals as they make up the majority of the test pool for calibration modeling and adjustments to the operation of commercial pulse oximeters. Therefore, commercial pulse oximeters are naturally biased to individuals with higher melanin contents (e.g., highly pigmented). Further, as illustrated in FIG.17, changing absorption over output range changes the effective emission wavelength.

[0092] FIG.18 further shows the shift in melanin induced peak wavelength shift and output curve distortion of an original LED and a melanin filtered LED. Signal ratio Red / IR is very sensitive to wavelength. Manufacturers use LED peak wavelengths that maximize Equation 4 which occurs where O2 absorption changes most rapidly with wavelength. (^^100%^^,red − ^^ 0%^^,red ) (Eq. 4)

[0093] Equation 5 is ato investigate the effect of melanin absorption on SpO2 accuracy using LED spectra. ^^^^0%0% ^^^^,IR−^^^^,red^^^^2=

[0094] Equation 6photoplethysmographic signal. The equation to the far right (e.g., integrals over wavelength) of Equation 6 models pulse oximetry of polychromatic sources. The far right side of Equation 6 produces the modeling results of FIGS.12, 20, and 21. Page 19 of 38 99391552.2PATENT Atty Docket: 093698-822591 Via Patent Center ^^ = (^^^^^^⁄ ^^^^ )red^^ art ^^,red∫^^^^^^,^^^^^^^^^(^^)^^(^^)^^^^ (^^^^^^⁄ ^^^^ )IR= ^^art^^,IR = ^∫^^^^^^,^^^^^^^^(^^)^^(^^)^^^^ (Eq.6)

[0095] FIG.19

[0096] FIG.20condition. As illustrated, melanin content results in higher SpO2 measurements than actual SaO2 measurements. The effect of melanin was calculated using the simple absorption only model of Equations 5 and 6 above. The error increases as SaO2 and SpO2 decrease, as illustrated in FIG.20.

[0097] FIG.21 illustrates a sensitivity analysis chart at SaO2 values of 90% using the model. Most commercial pulse oximeters have a FWHM bandwidth of about 15 nm to about 22 nm. As illustrated, using a peak wavelength of 660 nm and a FWHM bandwidth of 15 nm results in an error percentage of 0.5%. Using a peak wavelength of 660 nm and a FWHM bandwidth of 20 nm increases the error percentage to 4.7%. These results illustrate why some commercial pulse oximeters have significantly greater biases than others. The percentage error comes down to the FWHM bandwidth used in the light sources of the pulse oximeter.

[0098] In conclusion, analytical modeling results agree with empirical data from instruments when spectral power distribution is affected by melanin absorption. Accuracy of pulse oximeters is heavily dependent on spectral power distribution (wavelength, bandwidth). Most published results model pulse oximeters as monochromatic light sources, however, commercial pulse oximeters use polychromatic light sources. Varying skin pigmentation (melanin) affects the effective LED emission spectrum. Increasing melanin content leads to erroneously higher SpO2 readings. Error increases as SaO2 decreases. Errors can be made to match observed clinical discrepancies, depending on melanin concentration and LED characteristics. At 90-100% SaO2, there is a +1.5% error for highly pigmented individuals. At 70-80% SaO2, there is a 2.2% error for highly pigmented individuals.

[0099] FIG.22 illustrates the geometry used in modeling optical signals and detection for SpO2 readings. A simple hemispherical geometry is used to model the transmission of light and IR radiation through a fingertip. Pigmented epidermis was not included in the model because it is very thin compared to other distances involved and scatter and absorption that take place in these layers are assumed to be constant. Therefore, melanin layers only attenuate the signal before entering blood-perfused tissue and Page 20 of 38 99391552.2PATENT Atty Docket: 093698-822591 Via Patent Center then upon exiting. FIG. 23 illustrates the fundamental modeling equations. FIG. 24 illustrates the equations for solving for arterial oxygen saturation. FIG.25 illustrates IR light intensity and R light intensity in relation to SaO2, light attenuation, and time. Example 4:

[0100] Another experiment was conducted using three different pulse oximeters. The first pulse oximeter had a FWHM bandwidth of 22 nm, the second pulse oximeter had a 16 nm FWHM bandwidth, the third pulse oximeter included a bandpass filter having a 10 nm FWHM bandwidth. FIG.26 illustrates the emission light normalized spectral power and wavelength of the first pulse oximeter (i.e., 22 nm FWHM bandwidth). FIG. 27 illustrates the emission light normalized spectral power and wavelength of the second pulse oximeter (i.e., 16 nm FWHM bandwidth) and third pulse oximeter (i.e., 10 nm FWHM bandwidth). FIG.28 illustrates the relative bias or change in sensitivity in terms of bivariate slope and finger transmittance. The bivariate slope was significantly shallower for darker skinned subjects than the lighter skinned subjects, thereby indicating that 10 nm FWHM bandwidth of the third pulse oximeter significantly reduced the systemic skin pigmentation bias. The mean slope of the lighter skinned subjects was greater than that for the darker skinned subjects (one-tail Student t-test: P=0.009). Example 5:

[0101] Pulse oximetry uses noninvasive optical measurements of light transmission from each of two sources through vascularized living tissue over the cardiac cycle (SpO2). From those measurements, the relative amount of oxygenated hemoglobin (SaO2) in circulating blood can be deduced. Recent reports have shown that, compared with SaO2 measurements from blood samples, SpO2 measurements are biased erroneously high for patients with dark skin.

[0102] A new method, spectrally resolved photoplethysmography (srPPG), was developed to examine how spectral bandwidth affects the transmission of polychromatic light through the fingertip across the cardiac cycle. Spectral transmission was measured and recorded through the fingertip as the O2 concentration in inspired air was reduced. Digital spectral filters of two different bandwidths, narrow or broad, were applied to the same srPPG recordings to determine Page 21 of 38 99391552.2PATENT Atty Docket: 093698-822591 Via Patent Center whether SpO2 readings systematically varied for the two bandwidths. The srPPG method allowed measurement of the fractional amount of melanin in the optical path. The effect of melanin content on the ratio of SpO2 readings for narrow and broad spectral bandwidths was analyzed.

[0103] It was determined that the light emission spectra of light-emitting diode light sources, as used in commercial pulse oximeters, result in erroneously high SpO2 measurements for patients having greater melanin concentrations in their skin than those of the subject pool used for instrument calibration.

[0104] Pulse oximetry is based upon the principles of photo-plethysmography, which is an optical method incorporating one or more light sources and a photodetector. The light sources are directed to living vascular tissue and a photodetector receives a sampling of the light from either back-scattered (reflection mode) or forward-scattered (transmission mode) light. As blood volume increases and decreases across pulse cycles, there is a differential change in the constituent optical path lengths between arterial blood and all other media becoming longer and shorter, respectively.

[0105] With hypoxemia, the relative concentration of oxygenated hemoglobin (HbO2) decreases and the relative concentration of deoxygenated hemoglobin (Hb) increases. HbO2 transmits relatively more red light than infrared (IR) light and Hb transmits relatively more IR light than red light, making it possible to infer the relative concentrations of Hb and HbO2 from the pulse-dependent changes in optical path lengths of a red and an IR light. The specific peak wavelengths of the spectral power distributions of the red and the IR lights are chosen to maximize changes in transmission as a result of changes in concentrations of Hb and of HbO2. Typically, two light-emitting diode (LED) light sources are used in commercial pulse oximeters having peak emission wavelengths around 660 nm and 900 nm with full-width-half- maximum (FWHM) spectral bandwidths of approximately 20 nm and 50 nm, respectively.

[0106] As light passes through a fingertip, for example, IR light and red light transmissions change (ΔTIR and ΔTred) over each pulse cycle. Because factors not directly related to blood profusion (e.g. scatter, optical path length, melanin absorption) also affect transmission, the changes in transmission can be isolated more clearly if ΔTIR and ΔTred are each normalized to their mean transmission over pulse cycles. A Page 22 of 38 99391552.2PATENT Atty Docket: 093698-822591 Via Patent Center ratio of normalized transmission changes is defined and called R, which can be calculated using equation 7. ΔTred ^^ = ̅ T̅ ̅̅r ̅e̅ ̅̅dΔTIR (Eq.7) ̅^̅ ̅^̅^ ̅^^̅^

[0107] The value of R is predominately dependent on the absorption coefficients of blood for IR and red light. Ignoring wavelength-dependent optical scattering (i.e. Beer- Lambert absorption only), R is solely dependent on the two absorption coefficients in the limit as ΔT goes to zero. Equation 8 provides an alternate definition of R using logarithms. ^^ =Δlog (Tred) Δlog (TIR) (Eq.8)

[0108] This dependence of Rcoefficients is exact. Replacing T with its Beer-Lambert law expression where d is the optical path and α is the absorption coefficient for unit length is shown in Equation 9. ^^ = ^^−αd (Eq. 9)

[0109] Therefore, R equals the ratio of absorption coefficients for red and IR light as shown in Equation 10. ^^ =αred αIR (Eq.10)

[0110] Given a measurement of Rand Hb, the percentage of HbO2 can be calculated. For measurements of R to be useful to a clinician, the measurements must be converted to O2 saturation values. Equation 11 provides the conversion of R to O2 saturation valves (SaO2). ^^^^^^2 ≅ ^^^^^^2 = ^^1 − ^^2 ∗ ^^ (Eq. 11)

[0111] C1 and C2 are empirically determined constants of about 110 and 30, respectively.

[0112] The spectral transmission of melanin, a pigment that makes skin dark, increases with wavelength over the red and IR spectral range where pulse oximeter readings are taken. For polychromatic sources such as the LEDs used in commercial pulse oximeters, R changes systematically with melanin concentration in the skin because spectral transmittance changes exponentially with pigment density, according to the Beer-Lambert law; its effect on light transmission therefore does not cancel in the ratios defining R. In other words, transmission though a melanin filter Page 23 of 38 99391552.2PATENT Atty Docket: 093698-822591 Via Patent Center multiplicatively changes transmission wavelength by wavelength of the light reaching hemoglobin in the blood. For polychromatic light, this changes both the peak wavelength and the shape of the spectral power distribution of the light source, the former being more relevant to pulse oximetry.

[0113] R values are also specific to the specific optical properties of a given pulse oximeter design, so the values of R generated by any given pulse oximeter, (Equation 7), must be calibrated against direct SaO2 measurements. The U.S. Food and Drug Administration (USFDA) sets forth the pulse oximeter calibration method for all medical-grade or prescription oximeters in the USA. According to that method, a sample of no fewer than 12 individuals are subjected to controlled inhalation of air with different O2 concentrations while simultaneous SaO2 and R measurements are obtained. At fully saturated HbO2 concentration measured by SaO2, a given R value generated by a specific pulse oximeter is assigned an SpO2 value of 100%. As the measured SaO2 concentrations decrease, the device-specific R values are converted into equivalent SpO2 concentrations, (Equation 11).

[0114] The USFDA requires that at least 15% of the individuals sampled for the pulse oximeter calibration exercise be darkly pigmented, so individuals with dark skin have little influence on the resulting calibration of existing devices. Because of the systematic spectral shift in polychromatic light transmitted through melanin to longer wavelengths, the calibration constants in Equation 11 relating SpO2 to R will be different for light- and dark-skinned individuals. Consistent with the Beer-Lambert law and recent empirical reports, current pulse oximeters systematically overestimate O2 saturation among individuals with dark skin relative to individuals with light skin simply because they use polychromatic LED light sources. In clinical terms, this can mean that individuals with dark skin are less likely to be diagnosed with hypoxemia and therefore less likely to be treated for compromised respiratory function.

[0115] A study was conducted to demonstrate the effect of melanin concentration on R values generated by transmission mode pulse oximeters that use polychromatic light sources.

[0116] Eligible participants were healthy adults aged 18-60 years. Excluded from the study were those with a history of cardiovascular disease, pulmonary disease, anxiety, panic attacks, claustrophobia, and prescription medication use (except birth control), and subjects who were pregnant. Arterial blood pressure (BP), heart rate (HR), and an Page 24 of 38 99391552.2PATENT Atty Docket: 093698-822591 Via Patent Center SpO2 measurement near sea level were recorded prior to the start of the protocol to ensure values were within normal ranges, HR ≤ 120 beats min-1, BP 130 / 90 mm Hg, and SpO2 ≥ 95%.

[0117] Based on preliminary data for an analysis of statistical power, the study was planned to enroll 32 subjects as two groups, one with light skin and one with dark skin. Thirty-three subjects (18 females and 15 males, mean [range] age 36 [21-59] yr) were enrolled: 14 self-identified as White, 14 self-identified as Black, and five identified as neither Black nor White. None of the subjects were acclimated to living at high altitudes at the time of the study.

[0118] A Reduced Oxygen Breathing Device (ROBD2) was used to control, deliver, monitor, and record the percentage of O2 in the inspired gas mixture. The ROBD2 also continuously measured and recorded each subject’s SpO2 using a conventional pulse oximeter finger probe. Supplies of medical-grade compressed air, nitrogen, and O2were fed into the ROBD2, which controlled the flow rates of the gas components to supply O2-depleted air to the subject’s breathing mask while maintaining a constant average pressure of the gas mixture. The ROBD2 was programmed to step the subject through a series of progressively lower concentrations of O2 in the inspired gas mixture.

[0119] The basic approach used in photo-plethysmography was expanded to measure pulsatile changes in the spectral transmission of a polychromatic (i.e. white + near-IR) light through the fingertip, a technique called spectrally resolved photoplethysmography (srPPG). Measurements of srPPG provide pulsatile waveforms for a wide, continuous range of wavelengths almost simultaneously (within a few milliseconds).

[0120] The subject’s fingertip was inserted into a custom-built srPPG probe, between an LED lensed source and collection optics leading to a spectroradiometer. The LED source was a high-power broadband (white + near-IR) LED coupled with a collimating, total internal reflection LED lens to illuminate the bottom side of the fingertip. Transmitted light through the finger was sent through a right-angle prism and a fiberoptic collector to the entrance slit of a spectrometer. The spectrometer was connected to a 256 CCD camera to record the spectra at a frame rate of ~53.2 Sa seconds-1; 1500 frames of transmitted spectra were recorded over the wavelength range from ~610 to ~1000 nm. Every frame was stored for subsequent component Page 25 of 38 99391552.2PATENT Atty Docket: 093698-822591 Via Patent Center analysis and modelling. The entire process was automated such that every new acquisition file was obtained and stored immediately after the previous file had been stored. Thus, a nearly continuous srPPG data set was captured throughout an experimental session.

[0121] Skin reflectance spectra of the subjects’ fingers were measured using a halogen microscope ring light, and a fiberoptic cable (0.22 numerical aperture) located at the center of the ring and connected to a spectrophotometer. The light and fiber assembly were positioned ~35 mm from the subject’s finger. A spectrally neutral poly- tetrafluoroethylene rod was used as a 100% reflectance reference.

[0122] The procedures and the various devices used in the experiment were explained to every subject before the start of a session. The subject then washed their hands in warm water for 30-60 seconds to increase the temperature of their fingertips, thereby encouraging strong blood flow and enhanced perfusion. Subjects were fitted with the ROBD2 breathing mask, the ROBD2 SpO2 probe on the index finger of the right hand, and the custom-built srPPG probe on the middle finger of the right hand (Fig.29). The subject’s BP was recorded at the beginning and end of each session, but not during the procedure. The subject’s SpO2 and overall well-being were continuously monitored by a nurse throughout the session. A complete session comprised five step-wise changes in O2 concentrations (21%, 14%, 12%, 9%, and again, 21%) in the inspired air. At each stage of reduced O2 concentration, the subject was asked to report any symptoms they were experiencing and to confirm with the nurse whether they were comfortable to continue the experiment.

[0123] To achieve a stable baseline, the session started while the subject inspired air with 21% O2 for 1 min. Automatic srPPG data and ROBD2 data acquisitions were initialized during this baseline period. Once baseline data were collected and the subject was comfortable to proceed, the O2 concentration gradually decreased over 2 min to 14%, which was held at this concentration for 5 min before proceeding to the next stage. The inspired O2 was then reduced to 12% over 2 min and held for 5 min, then reduced to 9% over 2 min and held for no more than 5 min. The time spent at 9% O2depended on the nurse’s observation of the subject’s condition and the SpO2readings. If the subject was not in distress and their SpO2 readings were >75%, data acquisition continued for 5 min after which O2 concentration was returned to 21% over Page 26 of 38 99391552.2PATENT Atty Docket: 093698-822591 Via Patent Center 1 min. Data collection and monitoring continued for a few minutes after the subject recovered to baseline SpO2 concentrations.

[0124] The purpose of the study was to determine whether an interaction between the spectral bandwidth characteristics of light transmitted through the finger during SpO2 measurements and melanin concentration in the skin systematically affected measured R values. Using the srPPG data, it is possible to compare the effects of different spectral bandwidth characteristics to continuously determine R values over the duration of the experimental session for any subject. Based upon the Beer- Lambert law, melanin concentration should not affect SpO2 measurements if narrowband red and narrowband IR spectra were used in the determination of R values, but melanin concentration would affect R values if polychromatic spectra were used. Digital spectral bandpass filters were applied to the srPPG high-resolution spectral transmission data to make virtual red and IR sources having selected center wavelengths and bandwidths. R values, (Equation 7), were then determined for these virtual red and IR sources. The relationship between the polychromatic R values and the narrowband R values was analyzed by linear regression. The resulting slope (‘R- slope’) represents the bandwidth sensitivity of R. A unity R-slope value indicates that the narrowband sources are no different than those for the polychromatic sources and therefore has zero sensitivity to bandwidth. An R-slope value different than unity indicates sensitivity to bandwidth. If this sensitivity to bandwidth is systematically related to melanin concentration in the skin, SpO2 measurements will also systematically vary with melanin.

[0125] To determine whether there is a systematic relationship between the bandwidth sensitivity of R (R-slope) and skin pigmentation, it is necessary to quantify melanin concentration in the skin for different subjects. Unfiltered srPPG data sets were used to model the spectral transmission of the constituents in the fingertip. There are five major spectral absorbing components of the fingertip in the red to near-IR spectral region (610-1000 nm): Hb, HbO2, fat, water, and melanin. From published and widely used and accepted spectral transmission curves for the five constituents and the Beer-Lambert law, the relative contribution of melanin from the unfiltered srPPG spectral transmittance data sets was modeled. A wavelength-independent term was also used in the constituent modelling to account for individual differences in absolute Page 27 of 38 99391552.2PATENT Atty Docket: 093698-822591 Via Patent Center transmission as a result of optical coupling, scatter, and finger thickness. This method produced modelled transmission spectra that closely matched the measured spectra.

[0126] The fractional amounts of each constituent in a subject’s srPPG spectral transmission data set were determined by least-squares minimization of the difference between the five-constituent model and the subject’s srPPG spectral transmission curves over the range from 610 to 1000 nm in ~0.5 nm increments. Forty spectra were selected for analysis during the first 2.5 min of data collection, half of them occurring at plethysmograph waveform peaks and half at the troughs. The differences in five- constituent models between peak and trough estimations were naturally significant within subjects owing to changes in Hb and HbO2 over the cardiac cycle (paired t-test P=0.00016); however, the model differences were insignificantly small across subjects (non-paired t-test P=0.93). Therefore, the trough and peak measurements for a given subject were averaged. The five-constituent model amounts for Hb, HbO2, fat, water, and melanin were then determined for each subject. Melanin fraction is the amount of modelled melanin relative to the total molar concentration of all constituents without regard to the wavelength-independent term.

[0127] Skin reflectance spectra were used to evaluate skin color by the individual topography angle (ITA) metric, which divides skin color into six categories ranging from very light (ITA>55°) to dark (ITA<-30°). A correlation has been shown between melanin content and ITA. ITA was determined for a subset of 10 subjects of various skin colors (ITA range, -30.2 to 47.1) who were available midway through the recruiting period. ITA for each subject was based upon the average skin reflectance spectra from five fingers.

[0128] In total, 33 volunteer subjects were tested. Four subjects did not complete the protocol, but the truncated data for three of these subjects were included in the analyses. One subject completed two sessions because there was an equipment malfunction during the first session; only data from the second session were included in the analyses. Two subjects underwent the experimental procedure multiple times; the mean of the resulting R-slopes and melanin fractions were used in the analyses. R-slope values for two subjects, one of whom did not complete the entire session, were uncertain (r2<0.9) and their data were excluded from further analysis. Data from 31 subjects were analyzed. Page 28 of 38 99391552.2PATENT Atty Docket: 093698-822591 Via Patent Center

[0129] The bandwidths in the first condition (polychromatic) of the red and IR virtual sources were asymmetric Gaussian emissions designed to match the emission spectra of typical LEDs used in commercially available pulse oximeters. The center wavelengths and bandwidths were 662 nm with 20 nm FWHM for the red emitter and 902 nm with 51.5 nm FWHM for the IR. In the second (narrowband) condition, the same center wavelengths were used, but the spectral bandwidths and bandwidth shapes of the red and IR virtual sources were changed to 4 nm rectangular emissions, approximating a nearly monochromatic source such as a laser diode. These spectra are shown in FIG.30.

[0130] From the same srPPG spectral transmission data, R values were determined for the polychromatic sources and for the narrowband sources; the R-slope metric was determined for each subject. FIG. 31 shows the relationship between the virtual polychromatic R values and the virtual narrowband R values for one subject who had a high melanin concentration and identified himself as black.

[0131] The R-slope metric was found to correlate significantly with melanin fraction (R-slope, -20.8; Pearson’s correlation coefficient r=-0.647; P=0.00008). Subjects with higher melanin concentration had lower R-slope values (FIG.32).

[0132] The R-slope values were also found to differ significantly (FIG.33) between subjects with dark skin (fourth quartile melanin fraction) and subjects with light skin (first quartile melanin fraction).

[0133] Although most subjects with low melanin fractions had R-slope values close to unity (indicating minimal sensitivity to spectral bandwidth), the subjects with high melanin fractions had systematically lower slopes (indicating significant sensitivity to spectral bandwidth). The relationship between R-slope and melanin fraction shows that spectral bandwidth affects SpO2 measurements depending on the amount of melanin present. The experimental findings are consistent with observational data showing a bias in SpO2 measurements for darkly pigmented individuals and are grounded in a theoretical rationale for that bias. Specifically, R values, and therefore SpO2 measurements, are systematically affected by melanin concentration when polychromatic sources are used in pulse oximetry. Because R values obtained using current pulse oximeters are systematically higher for those patients with greater melanin pigmentation, SpO2 measurements in a clinical setting will systematically Page 29 of 38 99391552.2PATENT Atty Docket: 093698-822591 Via Patent Center overestimate SaO2 for more darkly pigmented individuals, particularly at low HbO2 concentration when clinical diagnosis and treatment are most critical.

[0134] The study clearly demonstrated the need for pulse oximeters using narrow spectral bandwidth red and IR sources. Importantly, the influence of skin pigmentation on R values cannot be eliminated simply by changing the demographics of the subjects used for SpO2 calibration. The USFDA requires only 15% of calibration subjects to be darkly pigmented. If a predominately darkly pigmented group were used for calibration instead, polychromatic pulse oximeters would more accurately measure SpO2 of darkly pigmented individuals, but those measurements would systematically underestimate the SpO2 of lightly pigmented individuals. In short, the skin pigmentation bias can only be eliminated by narrowing the spectral bandwidth of the light sources used in pulse oximetry. Pulse oximeters with narrow spectral bandwidths reduce and / or eliminate melanin bias.

[0135] Although the technique obviated SaO2 measurements to demonstrate the effect of spectral bandwidth, it is nevertheless important to show how SpO2 measurements, which are based upon R, are affected by spectral bandwidth. In particular, the calibration constants C1 and C2 will necessarily be different. Example 6:

[0136] Skin-color bias in pulse oximetry has been reported such that pulse oximetermeasurements (SpO2) overestimate the true arterial oxygen saturation (SaO2) ofindividuals with dark skin, leading to undiagnosed hypoxemia. This bias resulted from the optical properties of the two light-emitting diode (LED) light sources, one red and one infrared (IR), typically used in commercial pulse oximeters. The devices' detector measures the amounts of red light and IR light transmitted through the finger over each pulse cycle. Changes in red and IR light transmissions through arterial blood are isolated by normalizing them to their mean transmissions over the pulse cycle. The ratio of changes in normalized transmission (R) varies inversely with blood oxygenation.

[0137] To make pulse oximetry measurements meaningful to clinicians, R values are converted to SpO2values based on an empirical calibration relative to simultaneous SaO2 measurements. Skin-color bias arises from the polychromatic LED sources. These sources are not monochromatic, having full-width-at-half-maximum emissions of 20 to 60 nm. Based on the Beer-Lambert law, skin-color bias is caused by a melanin- Page 30 of 38 99391552.2PATENT Atty Docket: 093698-822591 Via Patent Center dependent wavelength shift of the transmitted polychromatic light spectral distributions; the greater the melanin concentration in the skin, the greater the wavelength shift. The study compared R values derived from narrowband (nearly monochromatic) sources and from polychromatic sources, showing that skin-color bias could be eliminated by using monochromatic red and IR light sources.

[0138] Some studies have suggested that, in addition to the overall melanin bias, there is a systematic relationship between skin colour and SpO2 measurement uncertainty, with uncertainty being greater for subjects with darker skin pigmentation. Although the Beer-Lambert Law can explain the source of melanin bias, it does not accurately model melanin distribution within the skin. Inferences based upon a simple application of the Beer-Lambert Law assume that spectrally absorbing substances, like melanin, are uniformly distributed throughout the medium, as if suspended in solution. In fact, discrete, spectrally absorbing melanin molecules (epidermal melanin units [EMUs]) are not evenly distributed throughout the epidermal layer of the skin. Moreover, the epidermal layer is not strictly planar but exhibits convoluted structures, as can be seen in images of stained skin sections published in literature. The discrete, non-uniformly distributed EMUs and textured boundaries of the epidermal layers strongly suggest that repeated SpO2 measurements from the same nominal area of the skin (e.g., fingertip) would not always be identical because the two light source emission-detection path lengths would vary. If this were true, the attenuation of the different optical path lengths through a greater number of EMUs in the epidermal layer would contribute to greater variability in SpO2 measurements.

[0139] Observed melanin bias in SpO2 measurements would necessarily be associated with measurement uncertainty. If true, then it would logically follow that the variability of independent measurements of SpO2 (based on R) would increase with overall melanin concentration in the skin, and also that measurement variability would be unrelated to R if using monochromatic light sources.

[0140] Eight of the 31 subjects from the study detailed in Example 5 were selected for repeated, independent R measurements. These subjects represented a range of skin pigmentation. As before, the subjects underwent a controlled oxygen desaturation procedure while data were continuously collected using spectrally resolved photoplethysmography (srPPG), as previously described. The procedure was performed on different days to permit determination of measurement uncertainty from Page 31 of 38 99391552.2PATENT Atty Docket: 093698-822591 Via Patent Center multiple independent measurements. Seven subjects completed three sessions, and one completed two sessions.

[0141] The techniques and metrics described herein were employed. The srPPG technique uses a custom fingertip oximeter with a broadband light source and a spectrometer to collect pulsatile waveforms for a wide, continuous range of transmitted wavelengths. To compare the effects of spectral bandwidth on the calculation of R, digital spectral bandpass filters were applied to the srPPG spectral transmission data. This method allowed for calculation of R for a narrowband light source and R for a polychromatic light source using the same spectral transmission data. The functional relationship between the two sets of R values (R-slope) for each experimental session was determined using a least squares linear regression. This R-slope metric represents the bandwidth sensitivity of the subject’s skin; an R-slope of unity indicates no bandwidth-related sensitivity, while a negative R-slope indicates a bandwidth- related sensitivity. R-slope was previously found to correlate with melanin content; the more melanin in the skin, the more negative the R-slope value.

[0142] To quantify the measurement-to-measurement uncertainty, the standard deviation of the R-slope measurements for every session (ŠR-slope) was calculated for each subject. Melanin fraction, a metric developed and described herein, was used to quantify the concentration of melanin in a subject’s finger. To determine melanin fraction, white light is transmitted through the fingertip and the spectral transmission data are collected. These data are then modelled using spectral transmission curves of the various constituents in the fingertip (water, bone, fat, melanin). Melanin fraction represents the amount of melanin needed in the model to predict the overall spectral transmission data, relative to the total of all the constituents. Melanin fraction correlates well with Individual Topography Angle, another method of estimating melanin based on skin spectral reflectance.

[0143] FIG. 34 illustrates the measurement-to-measurement uncertainty of the R- slope values (ŠR-slope) vs. melanin fraction for each subject. The standard deviation of the R-slope values (ŠR-slope) was found to correlate significantly with mean melanin fraction (Pearson’s correlation coefficient r=.837, P=0.0095). Subjects with higher melanin fraction had greater ŠR-slope values. Therefore, melanin concentration has the same effect on SpO2 measurement uncertainty when using polychromatic emitters as it does on measurement bias. These findings demonstrate that dark skin individuals Page 32 of 38 99391552.2PATENT Atty Docket: 093698-822591 Via Patent Center are doubly at risk for the misdiagnosis of clinically significant hypoxemia, firstly due to overall measurement bias and secondly due to measurement uncertainty.

[0144] The disclosures shown and described above are only examples. Even though numerous characteristics and advantages of the present technology have been set forth in the foregoing description, together with details of the structure and function of the present disclosure, the disclosure is illustrative only, and changes may be made in the detail, especially in matters of shape, size and arrangement of the parts within the principles of the present disclosure to the full extent indicated by the broad general meaning of the terms used in the attached claims. It will therefore be appreciated that the examples described above may be modified within the scope of the appended claims. Page 33 of 38 99391552.2

Claims

PATENT Atty Docket: 093698-822591 Via Patent Center CLAIMS What is claimed is:

1. A device for measuring oxygen saturation in blood, the device comprising: a first light source configured to emit a first light having a wavelength of about 630 nm to about 700 nm through a body part of a patient; a second light source configured to emit a second light having a wavelength of about 820 nm to about 1000 nm through the body part of the patient; a photodetector located either adjacent to the first light source and the second light source to operate in a reflectance mode or on an opposite side of the body part from the first light source and the second light source to operate in a transmittance mode, the photodetector configured to receive the first light and the second light; and a filter configured to narrow a full-width-half-maximum (FWHM) bandwidth of the first light and / or the second light to a narrow FWHM bandwidth, wherein the filter is a physical filter, an inherent filter to the first light source and / or second light source, or an electronic filter in communication with a spectroradiometer and processor, wherein the first light source and the second light source are rapidly counter phased.

2. The device of claim 1, wherein the filter narrows the FWHM bandwidth of the first light and / or the second light before being transmitted through the body part.

3. The device of claim 1, wherein the filter narrows the FWHM bandwidth of the first light and / or the second light received at the photodetector after being transmitted through the body part.

4. The device of claim 1, wherein the processor is configured to: receive data from the spectroradiometer; filter the data, via the electronic filter, to the narrow FWHM bandwidth; calculate an oxygen saturation value; and output the oxygen saturation value on a display in communication with the processor.

5. The device of claim 1, wherein the first light source and / or second light source is a laser diode or resonant-cavity LED and the inherent filter is a resonant cavity in the laser diode or resonant-cavity LED. Page 34 of 38 99391552.2PATENT Atty Docket: 093698-822591 Via Patent Center 6. The device of claim 1, wherein the physical filter is a narrowband interference filter.

7. The device of claim 1, wherein the narrow FWHM bandwidth is about 10 nm, about 5 nm, or less.

8. The device of claim 7, wherein the physical filter is directly deposited on the first light source and / or second light source or the physical filter is an encapsulate material configured to narrow the FWHM bandwidth.

9. The device of claim 1, wherein the first light source and the second light source comprise an LED, a laser diode, or combinations thereof.

10. The device of claim 1, wherein the device removes skin pigmentation bias from oxygen saturation measurements.

11. A method for measuring an oxygen saturation in blood, the method comprising: providing, via a first light source, a first light having a wavelength of about 630 nm to about 700 nm to a body part of a patient; providing, via a second light source, a second light having a wavelength of about 820 nm to about 1000 nm to the body part of the patient; receiving the first light and the second light at a photodetector located either adjacent to the first light source and second light source in a reflectance mode or located on an opposite side of the body part from the first light source and the second light source in a transmittance mode; filtering the first light and / or the second light to a narrow FWHM bandwidth; and measuring the oxygen saturation in blood of the patient based on the first light and the second light received at the photodetector, wherein the first light source and the second light source are rapidly counter phased.

12. The method of claim 11, wherein filtering the first light and / or the second light comprises using a filter before the first light and the second light are transmitted through the body part of the patient.

13. The method of claim 11, wherein filtering the first light and / or the second light comprises using a filter after the first light and the second light are transmitted through the body part of the patient. Page 35 of 38 99391552.2PATENT Atty Docket: 093698-822591 Via Patent Center 14. The method of claim 12 or 13, wherein the filter is a narrowband interference filter, an inherent filter to the first light source and / or second light source, or an electronic filter in communication with a processor.

15. The method of claim 14, wherein the first light source and / or second light source is a laser diode or resonant-cavity LED and the inherent filter is a resonant cavity in the laser diode or resonant-cavity LED configured to filter the first light and / or second light.

16. The method of claim 11, wherein the narrow FWHM bandwidth is about 10 nm, about 5 nm, or less.

17. The method of claim 11, wherein filtering the first light and / or second light comprises using a physical filter directly deposited on the first light source and / or second light source or the physical filter is an encapsulate material configured to narrow the FWHM bandwidth.

18. The method of claim 11, wherein the FWHM bandwidth removes skin pigmentation bias from oxygen saturation measurements.

19. A device for measuring oxygen saturation in blood, the device comprising: a first light source configured to emit a first light having a wavelength of about 630 nm to about 700 nm and a FWHM bandwidth of about 5 nm through a body part of a patient; a second light source configured to emit a second light having a wavelength of about 820 nm to about 1000 nm through a body part of a patient; and a photodetector located either adjacent to the first light source and the second light source in a reflectance mode or located on an opposite side of the body part from the first light source and the second light source in a transmittance mode, the photodetector configured to receive the first light and the second light, wherein the first light source is a resonant cavity LED, an LED having an interference filter, or a laser diode, wherein the first light source and the second light source are rapidly counter phased.

20. A device for measuring oxygen saturation in blood, the device comprising: a first light source configured to emit a first light having a wavelength of about 630 nm to about 700 nm through a body part of a patient; Page 36 of 38 99391552.2PATENT Atty Docket: 093698-822591 Via Patent Center a second light source configured to emit a second light having a wavelength of about 820 nm to about 1000 nm through a body part of a patient; a photodetector located either adjacent the first light source and the second light source in a reflectance mode or located on an opposite side of the body part from the first light source and the second light source in a transmittance mode, the photodetector configured to receive the first light and the second light; and a filter attached to the photodetector, the filter configured to narrow at least the first light received by the photodetector to a FWHM bandwidth of about 5 nm or less, wherein the first light source and the second light source are rapidly counter phased. Page 37 of 38 99391552.2

Images