Real-time photophysiology monitoring method and system

Abstract

A method of monitoring a subject using a photophysiological sensor system is described herein. The method comprises: obtaining a model of optical properties of at least one body tissue type to be monitored, where the model of optical properties comprises a definition of static (DC) and dynamic (AC) components of transmitted optical power and a definition of a source-detector interval related to a normalized path length of an illumination source for the photophysiological sensor; obtaining, from a wearable device worn by the subject, an indication of at least one physiological characteristic of the subject; obtaining an indication of at least one physical variable from a wearable device worn by the object; determining how the at least one physical variable affects the at least one physiological characteristic using the photophysiological model; and determining a corrected value of the physiological characteristic based on the determination of how the at least one physical variable affects the at least one physiological characteristic.

Description

Technical Field

[0001] The present disclosure relates to real-time photophysiological monitoring methods and systems. Background Art

[0002] Photoplethysmography (PPG) is a non-invasive optical technique used to monitor changes in blood volume or blood flow near the skin surface to determine various physiological parameters. The principle uses an illumination source and a photodetector to measure changes in the intensity of light as it passes through or reflects from body tissue. The detected optical signal is then analyzed and correlated with the pulsation of blood through the body stimulated by the heartbeat.

[0003] Photophysiological monitoring (beyond but including Beer-Lambert-based photoplethysmography, PPG) fits into the global mobile health market, which is growing rapidly in developed countries in response to aging populations and rising incidences of chronic diseases caused by lifestyle. Two main applications of the technology are: i) real-time and anytime clinician monitoring to capture vital health parameters of chronically ill subjects or those receiving post-operative care, and ii) self-monitoring and assessment, for example, for athletic performance, or to maintain or improve overall health and fitness levels. In addition, monitoring and assessment can be performed to identify or confirm potential diseases or monitor life-threatening parameters of subjects at risk to track potential conditions and provide early warning signs to prevent deterioration.

[0004] Of the current PPG systems, pulse oximetry devices (which determine blood oxygen saturation) are the most common. However, these devices are often inaccurate because they do not take into account the dynamic changes in tissue optical properties in living tissue, motion-induced artifacts, or the effects of sweat, skin creams, or sprays.

[0005] In the consumer fitness market, continuous heart rate monitoring systems based on electronic chest straps are commonly employed. However, these tend to be uncomfortable and conspicuous, making them unpopular for frequent use. Other wearable monitoring systems are emerging on the market, including watch-based sensors that can measure personal parameters such as heart rate, heart rate variability, oxygen saturation and / or respiratory rate. However, there are currently no wearable sensors that can simultaneously measure a wider range of these physiological parameters to meet clinical monitoring standards.

[0006] Therefore, the object of the present invention is to provide an opto-physiological (OP) sensor system, which includes 1) a multi-spectral illumination optical sensor (optical sensor), 2) customized electronics for driving the optical sensor and capturing signals from the optical sensor, 3) an embedded real-time artificial intelligence (AI) signal processing algorithm for generating physiological readings, a mobile or web application as a user interface, and 4) a method of monitoring an object using the optical sensor, which helps to solve the above-mentioned problems.

[0007] WO2015056007 describes an example of a method for assembling an optical physiological sensor. WO2021233319A1 describes a premature beat detection method, an electronic device, and a medium. WO2021213337A1 describes a method for monitoring the use of a wearable electronic device, a medium, and a wearable electronic device, and relates to the field of information processing technology. US20210393150A1 describes a device for non-invasively measuring biological information, comprising: a pulse wave sensor configured to measure a plurality of pulse wave signals with different wavelengths from an object; a contact pressure sensor configured to measure the contact pressure of an object while measuring the plurality of pulse wave signals; and a processor configured to obtain an oscillometric waveform based on the contact pressure and the plurality of pulse wave signals with different wavelengths, and obtain biological information based on the oscillometric waveform. US20220015716A1 describes an electronic device and method for sleep apnea detection. US20210330209A1 describes systems, devices, and methods for tracking one or more physiological indicators of a user (e.g., heart rate, blood oxygen saturation, etc.). US20210236014A1 describes a finger-wearable blood pressure monitor device, including a cuff, a tactile sensor array, and a control circuit. US20210196204A1 describes methods, systems, and methods for predicting the quality of sensor measurements. US20210375473A1 describes a wearable device that can be used for hypertension monitoring. US20210386363A1 describes an electronic device, including a translucent layer forming part of the exterior of the electronic device, an opaque material with limited microperforations positioned on the translucent layer, and a processing unit operable to determine information about the user via the translucent layer. US20200297226A1 describes an electronic fitness device, including a housing, a first optical transmitter array, a first optical receiver, and a second optical receiver. US20210212620A1 describes a mobile electronic device operable to be used in a vehicle and to determine a user's mental state (such as drowsiness). US20210353165A1 describes measuring a user's blood pressure using a wearable device, comprising extracting features associated with a pulse wave using sensor data of the wearable device; determining a pulse transit time (PTT); using at least one of the PTT scaled features to obtain a scaled feature; using the scaled feature as an input to a machine learning (ML) model; and using the output of the ML model to obtain the user's blood pressure. CN111588385A describes a method and device for determining oxygenated hemoglobin saturation. US20210121109A1 describes a method and system for a light emitting diode (LED) driving circuit for an optical probe. US20210345899A1 describes a multi-sensor patch for simultaneous abdominal monitoring of maternal and fetal physiological data, comprising a multi-layer flexible substrate.WO2021250224A1 describes a device and method for estimating one or more hemodynamic parameters (such as cardiac output or stroke volume). US20220000435A1 describes a method and device for determining respiratory information of a subject. WO2021101705A1 describes devices, systems, and techniques for monitoring the condition of a subject. US20210290060A1 describes systems and methods provided for remote object management and monitoring. WO2021194622A1 describes a method for measuring physiological characteristics of a user. US20210060343A1 describes a system and method for managing pain in a subject. US20210178164A1 describes a method and device for pain management using sleep detection.

[0008] Klodell CT et al., Oximetry-derived perfusion index for intraoperative identification of successful thoracic sympathectomy, Ann Thorac Surg 2005;80:467-70 described how an intraoperative PI derived from pulse oximetry was an additional indicator of successful thoracic sympathectomy in patients with upper limb hyperhidrosis.

[0009] Park J, Seok HS, Kim SS and Shin H (2022) Photoplethysmogram Analysis and Applications: An Integrative Review. 12: 808451 is a review paper that aims to examine existing research on photoplethysmogram, especially from an engineering perspective, regarding its generation mechanism, measurement principles, clinical applications, noise definition, preprocessing techniques, feature detection techniques, and post-processing techniques for photoplethysmogram processing. Summary of the invention

[0010] Various aspects of the invention are set out in the independent claims, and optional features are set out in the dependent claims. Various aspects of the invention may be provided in combination with each other, and several features of one aspect may be applied to other aspects.

[0011] In a first aspect, a method for monitoring an object using a photophysiological sensor system that takes into account physical factors is described. The photophysiological sensor system includes at least one sensor. The method includes:

[0012] obtaining a model of optical properties of at least one body tissue type to be monitored, wherein the model of optical properties includes a definition of a static component and a dynamic component of transmitted optical power and a definition of a source-detector spacing related to a normalized path length of an illumination source for the optical physiological sensor;

[0013] obtaining an indication of at least one physiological characteristic of the subject from a wearable device including at least one sensor worn by the subject;

[0014] obtaining an indication of at least one physical variable from a wearable device including at least one sensor worn by a subject;

[0015] determining, using the model, how at least one physical variable affects at least one physiological characteristic; and

[0016] A corrected value for the physiological characteristic is determined based on a determination of how the at least one physical variable affects the at least one physiological characteristic.

[0017] The at least one sensor may include at least one sensor configured to obtain an indication of at least one physiological characteristic and at least one sensor configured to obtain an indication of at least one physical variable or parameter (such as activity). The at least one sensor configured to obtain an indication of at least one physiological characteristic may be an optical sensor. The optical sensor may be a multispectral optical sensor capable of receiving electromagnetic radiation over a range of wavelengths.

[0018] The at least one sensor configured to obtain an indication of at least one physiological characteristic may be configured to obtain an indication of at least one of: heart rate (HR), perfusion index (PI), oxygen saturation (SpO2%), respiration rate (RR), blood pressure (BP), pulse rate variability (PRV), pulse transmitted time (PTT), and pulse wave velocity (PWV). For example, the at least one sensor configured to obtain an indication of at least one physical variable or parameter may be configured to obtain an indication of at least one of: contact pressure F c , contact temperature T A , acceleration a x,y,z , angular velocity Ω x,y,z , and absolute orientation ° x,y,z .

[0019] Obtaining a model of optical properties of at least one body tissue type to be monitored may include obtaining a model of photophysiological properties of at least one body tissue type to be monitored. Obtaining a model of optical properties of at least one body tissue type to be monitored may include obtaining a model of optical properties based on physiological variables of the individual. The physiological variables of the individual may include at least one of: body tissue type to be monitored, tissue / skin age, weight, presence of tattoos, tissue location, and / or skin color. In this way, the model can be used to accurately determine a corrected value of a physiological property for an individual regardless of their age, skin color, weight, and presence of tattoos.

[0020] Obtaining an indication of at least one physiological characteristic of the subject from a wearable device worn by the subject may include obtaining at least one physiological reading of the subject from the wearable device worn by the subject. In this case, determining using a model how at least one physical variable affects the at least one physiological characteristic may include determining using a model how at least one physical variable affects the at least one physiological reading.

[0021] The definition of the source-detector spacing associated with a normalized path length of the illumination source for the optical sensor may be a definition of the source-detector spacing associated with an optimized path length of the illumination source for the optical sensor.

[0022] The method may further include determining measured values ​​of precise physiological parameters using the corrected values ​​for the physiological characteristics, the physiological parameters including at least one of: heart rate (HR), perfusion index (PI), oxygen saturation (SpO2%), respiratory rate (RR), blood pressure (BP), pulse rate variability (PRV), pulse transit time (PTT) and pulse wave velocity (PWV).

[0023] The at least one physical variable may include at least one of the following: contact pressure F c , contact temperature T A , acceleration a x,y,z , angular velocity Ω x,y,z , and absolute orientation ° x,y,z .

[0024] The method may further include defining a minimum optical power dP that penetrates into a point (x, y) on the surface of the body tissue type according to Equation 1:

[0025] ρ(λ,l′(x′,y′)):dP(λ,x,y)=∫∫I0(λ,x′,y′)×e -ρ(λ,l′(x′,y′)) dx′dy′ [Equation 1]

[0026] where I0(λ,x,y) represents the total light entering through the entire surface of the body tissue type and I0(λ,x,y) undergoes an exponential decay that is a function of the optical density ρ(λ) of the body tissue type;

[0027] where the source-detector spacing is based on the s ,y s ) and an arbitrary detector at a point (x, y) on the surface of the body tissue type, such that and

[0028] where, for the through vectors x and y (from x _ Cross to x + and from y ) Cross to y + ) is a detector with a limited rectangular area, and the optical power received by the detector is according to equation 2:

[0029]

[0030] The model may include using the normalized physiological pulse function ψ(t) and the respective absorption coefficient μ according to Equation 3 ai (λ), scattering coefficient μ si (λ) and pulse coefficient μ pi (λ), defines the optical response of a dynamic and multi-layered body tissue type in terms of its dynamic optical density ρ(λ,l′,t), where the number of layers i ranges from 1 to N:

[0031]

[0032] The model may include separating the static components of Equation 3 according to Equations 4-6 and dynamic component

[0033]

[0034] The model may include defining the static component of the transmitted optical power for a rectangular detector defined by vectors x and y using the integral-sum rule for Equations 1 and 2 according to Equations 7, 8, and 9: and dynamic component like:

[0035]

[0036] The model may include defining the optimal source-detector spacing l'(λ) for an illumination source at wavelength λ according to Equation 10:

[0037]

[0038] The model may include assumptions of cylindrical symmetry and optical homogeneity of the body tissue type to be monitored, so that the optimal source-detector spacing l'(λ) is expressed as the position of the illumination source (x s ,y s ) or, conversely, a circle centered at the position of the detector (x d ,y d ) as the center.

[0039]

[0040] In some examples, the at least one physical variable includes acceleration a x,y,z , angular velocity Ω x,y,z and absolute orientation ° x,y,z At least one of, and wherein using the model to determine how at least one physical variable affects at least one physiological characteristic includes modeling the pumping action of the heart to determine the volumetric blood flow rate.

[0041] Modeling the pumping action of the heart may include determining the pressure gradient in the form of Equation 12:

[0042]

[0043] where A0 is the constant component of the blood pressure gradient from the heart, A1 is the amplitude of the fluctuating component, and ω p =2πf p , where f p is the pulse frequency, and the volumetric blood flow rate Q is determined by Equation 13 and Equation 14:

[0044]

[0045] Among them, ρ and μ f are the density and viscosity of the blood flowing through the blood vessels, and u z is the velocity of blood flow in the axial direction, and where the z-axis is taken along the axis of the arterial blood segment, and r is taken in the radial direction as a combination of acceleration, angular velocity, and absolute orientation that can be obtained when obtaining an indication of at least one physical variable from the wearable device.

[0046] In some examples, the at least one physical variable includes temperature, and wherein determining using the model how the at least one physical variable affects the at least one physiological characteristic includes modeling density of a particular tissue type as a function of wavelength and temperature of an illumination source of the optical physiological sensor.

[0047] Modeling the density of a particular tissue type as a function of the wavelength and temperature of the illumination source of the optical physiological sensor system may include modeling the density of the tissue according to Equation 15:

[0048]

[0049] where ρ(λ,t) indicates the density of a specific tissue type, C p is the tissue specific heat, V is the specific tissue volume, and A is the tissue surface area, such that A = πD 2 / 2. V and A depend on the average tissue diameter D = 2R, where R is the average radius of the tissue surface area A. The density and specific heat of blood are expressed by ρ b and Indicates that, while ω b is the volumetric blood flow rate. T A is the arterial temperature and is considered constant or equal to body temperature, while h 空气 is the heat transfer coefficient of the skin surface dominated by ambient heat exchange. T 空气 Indicates the ambient temperature.

[0050] In some examples, modeling the density of a particular tissue type as a function of the wavelength and temperature of an illumination source of the optical physiological sensor includes modeling the tissue temperature as a result of blood perfusion defined by thermoregulation of multiple layers using multi-wavelength illumination according to Equation 16:

[0051]

[0052] where k N , N , C pN ,ω bN , and q mN represents the thermal conductivity, density, specific heat, blood perfusion and metabolic heat production of the corresponding tissue layer (N), ρ b is the blood density, and represents the specific heat of blood, and T A is the body temperature. When the body temperature reaches equilibrium, for example, T A -T N =0, T N is considered a constant. In most cases, T A -T N > or <0, so Equation 16 represents the change in optical density ρ(λ,t).

[0053] In some examples, the at least one physical variable includes a contact pressure F between the optical physiological sensor and the object. c , and wherein using the model to determine how the at least one physical variable affects the at least one physiological characteristic comprises modeling changes in optical density ρ(λ,t) of a specific tissue type based on changes in optical path length l'(λ).

[0054] In some examples, determining using the model how at least one physical variable affects at least one physiological characteristic includes: when the contact force F c When within the selected range, the wavelength (λ) and temperature (T A -T N The optical density ρ(λ,t) as a function of the contact force F c proportional; and when the contact force F c When outside the selected range, the wavelength (λ) and temperature (T A -T N ) is modeled as a function of the contact force F c Proportional to the reciprocal of .

[0055] In some examples, the method may include modeling the change in optical path length l(λ,t) due to an increase or decrease in contact force as and The AC and DC components of the given optical density cause a change in optical density ρ(λ,t), such that

[0056]

[0057] In another aspect, a wearable optical physiological sensor system is provided, the wearable optical physiological sensor system being used to obtain physiological characteristics of a wearer and physical variables of the wearer's activities. The optical physiological sensor system comprises:

[0058] A photosensor panel comprising a plurality of photodiodes and a plurality of illumination sources (e.g., light emitting diodes), wherein at least one of the plurality of illumination sources (e.g., light emitting diodes) is selectively configured to emit light of a different wavelength than the other illumination sources, e.g., light emitting diodes;

[0059] Planar pressure sensor to obtain the contact force (F) between the optical physiological sensor and the wearer C ) instructions; and

[0060] A printed circuit board includes micro-electromechanical sensors for obtaining physical variables, a micro-control processor configured to process signals received from a plurality of micro-electromechanical sensors and a plurality of photodiodes, and a wireless communication interface.

[0061] The micro-electromechanical sensor may include at least two of a gyroscope, an accelerometer, and a digital compass.

[0062] The photosensor panel may be separated from the planar pressure sensor by a flexible substrate.

[0063] The planar pressure sensor can be separated from the printed circuit board by the flexible substrate.

[0064] The printed circuit board needs to be flexible.

[0065] The photosensor panel includes a plurality of temperature sensors configured to obtain an indication of the wearer's skin by contacting the wearer's skin.

[0066] The photosensor panel includes a plurality of temperature sensors equally spaced apart on the photosensor panel.

[0067] The processor may be configured to transmit raw sensor data or real-time processed data via the wireless communication interface.

[0068] The processor is configured to apply a machine learning model to sensor data obtained from the plurality of photodiodes, pressure data obtained from the planar pressure sensor, and a physical variable to generate modified output data, and to transmit the modified output data via a wireless communication interface.

[0069] The processor is configured to not obtain sensor data from the plurality of photodiodes or the micro-electromechanical sensor if the temperature sensor provides an indication of a temperature outside of a selected temperature range.

[0070] The microcontroller is configured to: provide the contact force (F) determined by the planar pressure sensor C ) range, no sensor data is obtained from the multiple photodiodes or microelectromechanical sensors.

[0071] In another aspect, a method for monitoring a subject using an optical physiological sensor system is provided. For example, such a method is particularly suitable when the subject is performing high-intensity exercise. The method includes:

[0072] obtaining a signal from a wearable device worn by a subject, the signal comprising a component indicative of at least one physiological characteristic of the subject and a component indicative of at least one physical parameter;

[0073] Modeling the obtained signal as a Taylor series based on an indication of at least one physical parameter, the Taylor series being a linear combination of the pulsatile signal and the motion term;

[0074] obtaining a corrected signal as a projection of the modeled pulsatile signal onto an orthogonal subspace; and

[0075] At least one physiological characteristic of the subject is determined using the corrected signal.

[0076] The movement of the sensor generates a small time-varying modulation of the received optical signal. Although small, it is comparable in size to the pulsation signal. Therefore, in a first approximation in small terms, the received optical signal consists of a DC component + a cardiac pulsation signal + a motion-related signal. In a first approximation, any product terms of the "motion-related" and "cardiac" contributions can be ignored.

[0077] The way in which this "motion" term relates to the values ​​measured by the triaxial accelerometer may depend on the effect of the elasticity of the tissue layers and the associated tissue strain on the absorption and scattering parameters; compressing the tissue increases the number of absorbers per unit path length traveled. For a roughly linear elastic tissue model, frequencies present in the measured accelerometer signal appear in the "motion-related signal" (with some amplitude and phase variation). It is these frequencies that are removed by the claimed method. We assume that this motion-related term can be derived from the measured accelerometer signal (a x 、a y and a z ) and its integrated velocity term (v x 、v y and v z ) with some weighted sum of weights N1 to N6. Including the velocity term allows some variation in amplitude and phase. Select N k The value of k = 1 ... 6 is such that the maximum signal power at the motion frequency is removed from the uncorrected signal at a selected time segment of the measured signal. This is equivalent to projecting onto an orthogonal subspace (with the six signals a x 、a y 、a z 、v x 、v y and v z The spanned subspace is the orthogonal subspace).

[0078] Obtaining the corrected pulsatile signal as the projection of the modeled pulsatile signal onto the orthogonal subspace is equivalent to M For an undetermined multiplier N k Minimize the integral shown in Equation 30, where s(t) is the obtained pulsating signal, a k Represents the measurement value of the three-axis accelerometer and its integrated speed

[0079]

[0080] Then, the corrected signal is

[0081]

[0082] Where N k The values ​​are those that minimize f(N).

[0083] The integral can be made over a length of T M The rolling time window can be, for example, T M = 8 seconds, refresh once per second.

[0084] The signal may include a pulsatile signal including an AC component and a DC component, and wherein obtaining the corrected pulsatile signal includes obtaining a projection of the measured AC signal onto an orthogonal subspace.

[0085] Large DC Contribution I DC It may be caused by light scattered by local tissue, while the AC component I AC (t) can be much smaller, and it includes the pulsation signal (or pulsation wave or pulsation waveform) and other time-related effects.

[0086] The method may further comprise filtering the pulsatile signal prior to minimizing the integral to remove any DC component, low frequency baseline offset and / or high frequency noise.

[0087] The method may further include filtering the signal with a zero-phase bandpass filter before obtaining the corrected ripple signal.

[0088] Obtaining a signal from a wearable device worn by the subject may include obtaining a plurality of pulsation signals from the wearable device, wherein each of the plurality of pulsation signals is a light signal of a different wavelength.

[0089] The method may further include performing an independent component analysis (ICA) or a principal component analysis (PCA) on each of the plurality of signals, optionally after the signals have been corrected. For example, motion may be first removed from all wavelengths and then principal component analysis may be performed on the remaining wavelengths. This may be better than performing PCA first, as the latter effectively only allows for a set of values ​​N k , while the former gives each wavelength its own set of N k The motion-related term is likely to depend on the wavelength and therefore appear different in the total reflected signal at that wavelength.

[0090] Obtaining an indication of at least one physical parameter from a wearable device worn by the subject may include receiving a plurality of signals indicative of the respective physical parameters from the wearable device, and normalizing the received plurality of signals between 0 and 1 to avoid any inconsistency in physical units.

[0091] In another aspect, a method for real-time monitoring of an object using an optical physiological sensor system is provided. The method facilitates rapid, real-time processing of data. The method comprises:

[0092] obtaining a pulsatile signal from an optical sensor that provides an indication of at least one physiological characteristic of the subject, the indication being from a wearable device worn by the subject;

[0093] obtaining an indication of at least one physical variable from a wearable device worn by the subject;

[0094] determining a dominant noise source in the pulsation signal based on the obtained indication of at least one physical variable;

[0095] determining a selected noise removal model from the list of selected models to be applied to the pulsatile signal; and

[0096] Using the selected noise removal model, a corrected value for the physiological characteristic is determined.

[0097] The list of selected models may include:

[0098] Orthogonal separation;

[0099] Acceleration- and / or balance-based models;

[0100] Temperature-based models; and

[0101] Model based on contact pressure.

[0102] Determining a dominant noise source in the pulsation signal based on the obtained indication of the at least one physical variable may include determining whether the obtained indication of the at least one physical variable exceeds a selected threshold, and wherein

[0103] Determining a selected noise removal model from the list of selected models to be applied to apply to the pulsation signal may include selecting the noise removal model based on a determination of whether the obtained indication of the at least one physical variable exceeds a selected threshold.

[0104] The method may further comprise: obtaining a model of optical properties of at least one body tissue type to be monitored, wherein the model of optical properties comprises a definition of a static component and a dynamic component of transmitted optical power and a definition of a source-detector spacing related to a normalized path length of an illumination source for the optical physiological sensor system; and

[0105] Wherein using the selected noise removal model to determine the corrected value for the physiological characteristic comprises using a model of photophysiological properties of at least one body tissue type and the selected noise removal model to determine the corrected value for the physiological characteristic.

[0106] The physical variable includes at least one of: contact pressure, temperature, acceleration, angular velocity, and absolute orientation.

[0107] In another aspect, a method for training a machine learning model for monitoring an object using a wearable device including an optical physiological sensor system is provided. The method comprises:

[0108] obtaining a pulsatile signal providing an indication of at least one physiological characteristic of the subject;

[0109] obtaining an indication of at least one physical variable as a function of time from the wearable device;

[0110] transmitting data representing the pulsation signal and the indication of the physical variable to a remote computing platform (i.e., a workstation or a computing cloud) via a wireless interface;

[0111] At the remote computing platform, training a machine learning model based on the received data; and

[0112] After training, the coefficients / parameters are sent from the remote computing platform to the embedded machine learning model in the wearable device.

[0113] At the remote computing platform, training the machine learning model based on the received data may include training a generative adversarial network (GAN), the GAN including two generators and a discriminator. One of the two generators is configured to learn an anti-noise mapping from an input noise pulsation signal to a reference signal, and the other of the two generators is configured to learn a noise distribution in a physical variable captured from the wearable device.

[0114] In another aspect, a method for training a machine learning model for monitoring an object using a wearable device including a light physiological sensor system is provided. The method comprises:

[0115] obtaining a pulsatile signal, the pulsatile signal providing an indication of at least one physiological characteristic of the subject;

[0116] obtaining an indication of at least one physical variable as a function of time from the wearable device;

[0117] processing the pulsatile signal and the indication of at least one physical variable using a machine learning model to provide an output indicative of a physiological characteristic;

[0118] transmitting data representing the pulsation signal and the indication of the physical variable to a remote computing platform via a wireless interface;

[0119] At the remote computing platform, training an improved machine learning model based on the received data;

[0120] Sending improved, trained machine learning models to wearable devices; and

[0121] The pulsatile signal is processed using the improved, trained machine learning model and the indication of at least one physical variable is processed using the machine learning model to provide an output indicative of a physiological characteristic.

[0122] While a user of the wearable device is performing a selected exercise routine, obtaining a pulsation signal from the wearable device that provides an indication of at least one physiological characteristic of the subject, and obtaining an indication of at least one physical variable as a function of time may be performed, and wherein a machine learning model is trained at a remote computing platform based on the exercise routine.

[0123] In another aspect, a wearable optical physiological sensor system is provided, which is used to obtain physiological characteristics of a wearer and physical variables of the wearer's activities. The optical physiological sensor includes:

[0124] A photosensor panel comprising a plurality of photodiodes and a plurality of illumination sources (e.g., light emitting diodes), wherein at least one of the plurality of illumination sources (e.g., light emitting diodes) is selectively configured to emit light of a different wavelength than the other illumination sources; and

[0125] a printed circuit board including a micro-electromechanical sensor for obtaining a physical variable, a processor configured to process signals received from a plurality of micro-electromechanical sensors and a plurality of photodiodes, and a wireless communication interface;

[0126] Therein, the plurality of illumination sources are located at intersections between a plurality of concentric circles, wherein each of the plurality of concentric circles is centered on a corresponding one of the plurality of photodiodes.

[0127] The plurality of illumination sources includes at least four different illumination sources, each illumination source being configured to emit light of a different wavelength. Four or more wavelength illumination sources provide a better choice because two wavelength illuminations are required to calculate oxygen saturation (SpO2%), and the responses of the plurality of photodiodes from the different wavelength illuminations can be considered for cross-correlation of each wavelength illumination to determine the best pulsatile signal obtained from the device.

[0128] The plurality of illumination sources includes at least sixteen different illumination sources, each illumination source configured to emit light of one of four or more different wavelengths, wherein the at least sixteen different illumination sources are grouped into four groups, each group configured to emit light of one of four or more different wavelengths.

[0129] In some examples, the shortest wavelength illumination source is located at the intersection of inner circles of a plurality of concentric circles, and wherein the longer wavelength illumination source is located at the intersection of concentric circles further outward from the center of the concentric circles, such that the longer wavelength illumination source is positioned further away from each of the photodiodes, and the wavelength of the illumination source increases with radial distance from each photodiode.

[0130] In some examples, the photosensor panel includes one, two, and three photodiodes.

[0131] At least one wavelength of the illumination source is 465 nm.

[0132] The optical physiological sensor further comprises at least one planar pressure sensor for obtaining a contact force (F) between the optical physiological sensor and the wearer. C ) instructions.

[0133] In another aspect, a method of obtaining an oxygen saturation (SpO2%) biomarker of a subject is provided, the method comprising:

[0134] irradiating the object with at least four different wavelengths λ1, λ2, λ3, λ4 to obtain a pulsating signal at each of the four different wavelengths λ1, λ2, λ3, λ4;

[0135] For four different wavelengths, the ratios of the height of the pulsatile signal to the associated DC background level, r(λ), are obtained and used in pairs to obtain the oxygen saturation biomarker of the subject.

[0136] The wavelengths are chosen to satisfy the condition for certain A, B, C, D:

[0137] ρ=Aσ(λ1,λ2)+B=Cσ(λ3,λ4)+D [Equation 32]

[0138] where σ(λ1,λ2)=r(λ1) / r(λ2); for the chosen wavelength pairing, the current device satisfies the condition well, as well as the condition implied by the most current oxygen saturation measurements.

[0139] For four different wavelengths, the ratio r(λ) of the height of the pulsatile signal to the associated DC background level is obtained, which can be obtained using a wearable opto-physiological sensor as described above and / or using other sensors such as a camera.

[0140] The value of p obtained from Equation 32 enables fitting of the oxygen saturation biomarker to a general (wavelength-independent) curve according to Equation 33

[0141] SpO2=SpO2(ρ)=α+βρ+γρ 2 …, [Equation 33]

[0142] During this period, the satisfaction of Equation 32 becomes an important step in validating this biomarker.

[0143] In another aspect, there is a computer-readable non-transitory storage medium comprising a program for a computer, the program being configured to cause a processor to perform a method according to any of the above aspects. BRIEF DESCRIPTION OF THE DRAWINGS

[0144] Embodiments of the present disclosure will be described, by way of example only, with reference to the accompanying drawings, in which:

[0145] Figure 1 An example of an optical physiological sensor system is shown;

[0146] Figure 2 Another example of an optical physiological sensor system is shown;

[0147] Figure 3 Another example of an optical physiological sensor system is shown;

[0148] Figure 4 A functional schematic diagram of an optoelectronic hardware system 400 is shown;

[0149] Figure 5 A functional schematic diagram showing an example method of processing a signal such as Figures 1 to 3 Photoelectric system or Figure 4 The optoelectronic hardware system is obtained by the optoelectronic sensor;

[0150] Figure 6 shows a pulsatility waveform of a subject with skin type I (Fitzpatrick scale) collected from the back of the left wrist using a photophysiological monitoring system;

[0151] Figure 7 shows a pulsatility waveform of a subject with skin type II (Fitzpatrick scale) collected from the back of the left wrist using a photophysiological monitoring system;

[0152] Figure 8 shows a pulsatility waveform of a subject with type III skin (Fitzpatrick scale) collected from the back of the left wrist using a photophysiological monitoring system;

[0153] Fig. 9 shows a pulsatile waveform of a subject with skin type IV (Fitzpatrick scale) collected from the back of the left wrist using a photophysiological monitoring system;

[0154] Fig.10 shows the pulsatile waveform of a subject with type V skin (Fitzpatrick scale) collected from the back of the left wrist using a photophysiological monitoring system;

[0155] Fig.11 shows a pulsatile waveform of a subject with skin type VI (Fitzpatrick scale) collected from the back of the left wrist using a photophysiological monitoring system;

[0156] Fig.12 shows the pulsation waveform of a male subject's skin type VI (Fitzpatrick scale) monitored using 4 wavelengths of light physiological monitoring; and

[0157] Fig.13 Shown are pulsation waveforms of a male subject's skin type VI (Fitzpatrick scale) monitored using 4 wavelengths of light physiological monitoring. DETAILED DESCRIPTION

[0158] The following symbols will be used throughout this description:

[0159] μ a Absorption coefficient

[0160] μ s Scattering coefficient

[0161] Refractive index

[0162] ρ(λ) optical density

[0163] Path length

[0164] I(λ) Light intensity

[0165] P Optical power

[0166] μ f Blood viscosity

[0167] u z Blood speed

[0168] Q Volume blood flow rate

[0169] C p Tissue specific heat

[0170] V Specific tissue volume

[0171] A Tissue surface area

[0172] T A Arterial temperature

[0173] h 空气 Heat transfer coefficient

[0174] T 空气 Ambient air temperature

[0175] k Thermal conductivity

[0176] q m Metabolic heat production

[0177] g Anisotropy factor

[0178] F c Contact Force

[0179] ∈ Extinction coefficient

[0180] C absorber concentration

[0181] Proxy parameter

[0182] T M Measurement time interval

[0183] a x,y,z Acceleration

[0184] Ω x,y,z Angular velocity

[0185] ° x,y,z Absolute orientation

[0186] Figure 1 An example photophysiological sensor system 100 is shown for monitoring or determining a photophysiological property of a subject. The photophysiological sensor system 100 may be designed to be wearable, such as a patch, a watch, or a ring. Figure 2 Another example of an optical physiological sensor system 200 is shown, and Figure 3 Another example of a photophysiological sensor system 300 is shown.

[0187] As in Figure 1 As can be seen in , the photophysiological sensor system 100 may include: one or more sensors (including, for example, multi-spectral / wavelength sensors), electronics for controlling the sensors and receiving signals from the sensors, and a computer program running on a processor, the computer program including an algorithm configured to process the signals, for example, according to a photophysiological model described in more detail below. For example, the photophysiological sensor system 100 may include a photodiode 101, a temperature sensor 103, and a plurality of illumination sources 105, 107, 109, 111, each illumination source being configured to emit light of a different corresponding wavelength. It may also include a MEMS motion sensor 135.

[0188] The optical physiological sensor system 100 is configured to obtain at least one physiological reading of the subject, and the at least one physiological reading can provide an indication of at least one physiological characteristic of the subject. The optical physiological sensor system 100 can also be configured to obtain an indication of at least one physical variable related to the wearer's activities. These activities can then be considered in the optical physiological model to obtain more accurate optical physiological measurements, such as heart rate (HR), perfusion index (PI), oxygen saturation (SpO2%), respiratory rate (RR), blood pressure (BP), pulse rate variability (PRV), pulse transit time (PTT) and pulse wave velocity (PWV).

[0189] exist Figure 1 In the example shown, the photophysiological sensor system 100 includes a photosensor panel 121, a plurality of illumination sources 105-111, and optionally at least one temperature sensor 103 (in Figures 1 to 3The photoelectric sensor panel 121 includes three photodiodes 101 (although there are four in the example of Figure 2 The example shown includes two photodiodes 101 and Figure 3 The example shown includes only one photodiode 101), and the plurality of illumination sources 105-111 are illumination sources (in Figure 1 In the example, there are 24 Figure 2 In the example, there are 24, and Figure 3 In the example, there are 16).

[0190] The optical physiological sensor system 100 further includes: a planar pressure sensor 123 and a flexible printed circuit board 125. The planar pressure sensor 123 is used to obtain the contact force (F) between the optical physiological sensor and the wearer. C ) indication. The flexible printed circuit board 125 includes: a micro-electromechanical sensor 135 for obtaining a physical variable; a micro-control processor 405, configured to process signals received from a plurality of micro-electromechanical sensors and a plurality of photodiodes 101. In some examples, the optical physiological sensor system 100 may also include an optional wireless communication interface 413, for example, located on the printed circuit board 125, as described below with reference to Figure 4 The micro-electromechanical sensor and the micro-control processor may be located on a real-time signal processing circuit board 131 mounted on the printed circuit board 125. The real-time signal processing circuit board 131 may include a computer program and / or instructions, which means that the real-time signal processing circuit board 131 is configured to receive and process sensor signals according to the photophysiological model, as described in more detail below.

[0191] In this manner, the micro-electromechanical sensors 135, 401 and the micro-control processor are located on the opposite side of the photosensor from the photodiode, temperature sensor and illumination source.

[0192] In the example shown, the micro-electromechanical sensor 135, 401 includes a 3-axis gyroscope, a 3-axis accelerometer, and a 3-axis digital compass. Figure 1 As shown, the photosensor panel 121 is separated from the planar pressure sensor 123 by the first flexible substrate 127 , and the planar pressure sensor 123 is separated from the printed circuit board 125 by the second flexible substrate 129 .

[0193] The photosensor panel 121 includes a plurality of temperature sensors 103 (four in this example) configured to obtain an indication of the wearer's skin by contacting the wearer's skin. The plurality of temperature sensors 103 are equally spaced apart on the photosensor panel 121.

[0194] At least one of the plurality of illumination sources 105-111 is optionally configured to emit light of a different wavelength than the other illumination sources. The wavelength may be, for example, 525 nm, 595 nm, 650 nm, 910 nm, or others. As discussed in more detail below, using multiple (in this example, four) different wavelengths may better accommodate different tissue types, tissue locations, and skin tones. In some examples, a wavelength of 475 nm may be preferred.

[0195] exist Figure 1 and Figure 2 In the example of , there are six LEDs for each wavelength, and there are four or more different wavelengths, which means a total of 24 or more LEDs. Figure 1 and Figure 2 As shown, the light emitting diodes 105-111 are located at the intersection of a plurality of concentric circles, wherein each of the plurality of concentric circles is centered on a corresponding one of the plurality of photodiodes 101. Figure 3 In the example of FIG. 1 , LEDs 105 - 111 are equally spaced around the circumference of a series of concentric circles centered on photodiode 101 , but in a manner such that LEDs in adjacent circles are offset by 90°, thereby increasing the spacing between illumination sources (eg, LEDs in adjacent circles).

[0196] exist Figure 1 and Figure 2 In the example of , the light emitting diode of the shortest wavelength (wavelength 1) is located at the intersection of the inner circles of the plurality of concentric circles, and the illumination source, for example, the light emitting diode of a longer wavelength (wavelength 2 to 4, where 4 is the longest wavelength and 1 is the shortest wavelength), is located at the intersection of the concentric circles further outward from the center thereof, so that the illumination source (for example, the light emitting diodes 105 to 111 of the longer wavelength) is located further away from each photodiode, and the wavelength of the illumination source (for example, the light emitting diode) increases with the radial distance from each photodiode 101. In this way, in Figure 1 and Figure 2 In the example, the wavelength 1 LED is on the inside, followed by the wavelength 2 LED, then the wavelength 3 LED, and finally the wavelength 4 LED is on the outside. Figure 3 In the example of , the illumination sources (e.g., light emitting diodes) are simply arranged so that longer wavelength light emitting diodes are positioned further away from the photodiode 101, and the wavelength of the illumination sources (e.g., light emitting diodes) increases with radial distance from the photodiode 101. Preferably, at least one of the wavelengths is blue light, for example at 465 nm, as it has surprisingly been found that blue light is particularly effective when used to determine SpO2 for specific tissue types and locations.

[0197] The microcontroller (MCU) 405 is configured to process sensor data obtained from the plurality of photodiodes 101, pressure data obtained from the planar pressure sensor 123, and physical variables from the MEMS motion sensor 135 by applying a photophysiological model to the data (as described in more detail below) to obtain an indication of a physiological characteristic of the subject / wearer of the photosensor 100.

[0198] In some examples, the microcontroller 405 can be configured to implement data processing by transmitting raw sensor data or real-time processed data via a wireless communication interface. The microcontroller 405 can be configured to apply a machine learning model to the sensor data obtained from the plurality of photodiodes 101, the pressure data obtained from the planar pressure sensor 123, and the physical variable to generate modified output data, and send the modified output data via the wireless communication interface 413.

[0199] In some examples, the microcontroller 405 is configured to not obtain sensor data from the plurality of photodiodes 101 or the microelectromechanical sensor 135 when the one or more temperature sensors 103 provide an indication that the temperature (T) on the skin / organ tissue of the contacted object is outside the selected temperature range of 18° C.-37° C. Additionally or alternatively, in some examples, the microcontroller 405 is configured to not obtain sensor data from the plurality of photodiodes or the microelectromechanical sensor when the planar pressure sensor 123 provides an indication of a contact pressure outside the determined pressure range.

[0200] When the wavelength 1 LED is fully illuminated, one or more photodiodes begin to capture the backlight from the contacting skin tissue, and then the wavelength 2 LED is turned on to capture the backlight again, the same as the previous process. The wavelengths 3 and 4 LEDs perform the same process as described. One complete duty cycle of backlight acquisition of the four wavelength illumination sources (i.e., LEDs plus the background signal captured by the same photodiode) is set as one (1) sampling point consisting of 5 separate values. Therefore, the effect of simultaneous sampling of multiple wavelengths is achieved as expected. There are five (5) channel signals of four (4) wavelength illuminations plus one (1) background captured by one or more photodiodes of the multi-wavelength illumination photosensor.

[0201] As described above, in order to process the data received by the photoelectric sensor 100 and obtain useful physiological parameters of the subject, a photophysiological model is applied. In related art systems, the Beer-Lambert model has been applied to relate the absorption of light to the properties of the material through which the light travels. However, it has been found that this approach is too basic to be truly representative when attempting to establish the photophysiological properties of body tissue, and therefore a photophysiological model that takes into account the various properties of the body tissue to be monitored is needed.

[0202] In the present case, the photophysiological model includes the definition of static and dynamic components of the transmitted optical power and the definition of the source-detector separation related to the normalized path length of the illumination source (eg, light emitting diode) of the photophysiological sensor.

[0203] The at least one physical variable includes at least one of: contact pressure Fc, temperature, acceleration, angular velocity, and absolute orientation.

[0204] In this model, light transmission through the tissue medium can be represented as an exponential decay according to Eq. 1, .

[0205] I(λ)=I0(λ)×e -ρ(λ) [Equation 1]

[0206] This describes the exponential decay of the intensity I0(λ) of light of wavelength λ as it passes through a tissue medium with an optical density ρ(λ).

[0207] The model includes defining the minimum optical power dP that penetrates to a point (x, y) on the surface of the body tissue type according to Equation 2:

[0208] ρ(λ,l′(x′,y′)):dP(λ,x,y)=∫∫I0(λ,x′,y′)×e -ρ(λ,l′(x′,y′)) dx′dy′ [Equation 2]

[0209] where I0(λ,x,y) represents the total light entering through the entire surface of the body tissue type and undergoes an exponential decay that is a function of the optical density ρ(λ) of the body tissue type;

[0210] The source-detector spacing is based on the (x s ,y s ) and the light source at The path length is defined by an arbitrary detector at a point (x, y) on the surface of a body tissue type, such that and

[0211] where, for the through vectors x and y (from x - Cross to x + and from y - Cross to y +) is a detector with a finite rectangular area defined by the light source, the optical power received by the detector is according to equation 3:

[0212]

[0213] Therefore, the model consists in using the normalized physiological pulse function ψ(t) according to Equation 4 and the absorption coefficient μ respectively ai (λ), scattering coefficient μ si (λ) and pulse coefficient μ pi (λ), and μ pi (λ), defines the optical response of a dynamic and multi-layered body tissue type in terms of its dynamic optical density ρ(λ,l′,t), where the number of layers i ranges from 1 to N:

[0214]

[0215] The model consists of separating the static components according to Equations 5 to 6. and dynamic component As the DC component and AC component of Equation 4 respectively:

[0216]

[0217] The model involves using the integral-sum rule for Equations 2 and 3 to define the static component of the transmitted optical power of the rectangular detector defined by vectors x and y according to Equations 7, 8, and 9: and dynamic component like:

[0218]

[0219] For a source at wavelength λ, the model consists in defining the optimal source-detector spacing l related to the normalized path length according to Eq. 10 ′ (λ):

[0220]

[0221] The model includes the assumption of cylindrical symmetry and optical homogeneity of the body tissue type to be monitored, so that the optimal source-detector spacing l is related to the normalized path length according to Eq. ′ (λ) is expressed as the position of the light source (x s ,y s ) as the center, or conversely, the position of the detector (x d ,y d ) as the center:

[0222]

[0223] During physical intensity (PI) activities, the optical properties of tissues are also affected by acceleration (m 2 / s), angular velocity (° / s), body temperature regulation (℃), and contact force (N / m 2 )’s control or influence.

[0224] The pumping action of the heart can be measured using a pressure gradient To express as shown in formula 12:

[0225]

[0226] Where A0 is the constant component of the blood pressure gradient from the heart, A1 is the amplitude of the fluctuating component (e.g. in the form of systolic and diastolic pressures), ω p =2πf p and f as the pulse frequency p .

[0227] In this paper, the arterial blood wall is considered to be rigid, the blood flow is laminar, there is rotational symmetry of the flow, and the velocity changes along the length of the vessel are small compared to the rate of change of velocity with respect to time, and then the velocity is calculated in the vessel (i.e., as a cylindrical, polar coordinate Arteries)

[0228]

[0229] Among them, ρ and μ f are the density and viscosity of blood flowing through the blood vessels, and u z is the velocity of blood flow in the axial direction (angle). Therefore, these two physical states of blood flow, such as laminar flow and angular direction, explain the factors regulating the acceleration and balance of blood flow during physical activities.

[0230] The volumetric blood flow rate Q (in the transformed space) can be determined as:

[0231]

[0232] Therein, the z-axis is taken along the axis of the arterial blood segment and r is taken in the radial direction as a combination of acceleration and balance angle that can be determined and captured by a 3-axis accelerometer and a 3-axis gyroscope.

[0233] Under the same mechanism combination of laminar flow and equilibrium angle, Q can also diffuse into capillaries or peripheral blood vessels that cross-connect with the tissue body. These dynamic contraction and relaxation behaviors can be reflected in changes in the optical properties of the tissue, such as the absorption coefficient (μ a ), scattering coefficient (μ s ), refractive index (n), and optical density ρ (λ).

[0234] Since capillaries or peripheral blood flow is controlled by main arteries or branch arteries, the content and density of these capillaries or peripheral blood changes with the changes in blood flow from the local or even the entire cardiac system. These phenomena can be captured or detected by multi-wavelength illumination photoelectric sensors, such as the amplitude changes corresponding to different wavelength illuminations and the dynamic contraction and relaxation behaviors.

[0235] Tissue temperature (°C) is a direct result of defined blood perfusion as thermoregulation, reflecting variations from tissue composition, skin thickness, surface area, tissue volume and ambient temperature in the presence of living tissue properties or body tissue environment.

[0236] For a specific tissue type, this phenomenon can be expressed as Equation 15:

[0237]

[0238] where ρ(λ,t) indicates the density of a specific tissue type, C p is the tissue specific heat, V is the specific tissue volume, and A is the tissue surface area, such that A = πD 2 / 2.

[0239] V and A depend on the average tissue diameter D = 2R, where R is the average radius of the tissue surface area A. The density and specific heat of blood are expressed by ρ b and Indicates that, while ω b is the volumetric blood flow. T A is the arterial temperature and is considered constant or equal to body temperature, while h 空气 is the heat transfer coefficient of the skin surface dominated by ambient heat exchange. T 空气 Indicates the ambient temperature.

[0240] For multiple layers with multi-wavelength illumination:

[0241]

[0242] where k N , N , C pN ,ω bN , and q mN represents the thermal conductivity, density, specific heat, blood perfusion, and metabolic heat production of the corresponding tissue layer (N). In this equation, ρ b is the blood density, and represents the specific heat of blood. A is body temperature and is considered a constant.

[0243] Thermoregulation presents body temperature changes that are not only dominated by the main or majority arterial blood flow, but also extend to local tissue temperature changes. Either way, local tissue temperature changes can simulate thermoregulation to affect cardiac blood flow that reflects body temperature changes. In the reality of photophysiological monitoring, the amplitude of the pulsation waveform or pulse wave can be detected, which clearly indicates the absorption coefficient (μ a ), scattering coefficient (μ s ), refractive index (n), and optical density ρ(λ,t).

[0244] Contact force as contact pressure (F C ) is a physical variable or factor that affects the amplitude of the generated signal, and thus the signal-to-noise ratio of the detection. Photophysiological monitoring shows that excessive or insufficient contact pressure can cause changes in the detected pulsation waveform or pulse wave. As the contact force (F) between the sensor and the skin increases, the C ) increases, the DC amplitude increases, while the AC component increases first. At high contact forces (F) between the optical physiological sensor and human skin, C ) shows the increase of the AC and DC components of the signal until a maximum range (designated as the optimum contact pressure). As the pressure increases beyond this range, the pulsating waveform is no longer recognizable.

[0245] Referring to [Equation 4-6], the change in optical path length caused by the increase or decrease in contact force causes and The variation of the optical density ρ(λ,t) in terms of the AC and DC components is given by

[0246]

[0247] The pulse amplitude peaks at different contact forces, indicating that individual differences cause variations related to the actual force applied to the arterial wall. The peak pulse amplitude ranges from, for example, 0.15 N to 1.5 N. The sensor system picks up the contact force signals and processes them individually with an AI-driven training strategy.

[0248] The wavelength-dependent optical properties of the tissue change depending on the contact pressure value. The deformation of the tissue structure due to the contact pressure affects the scattering properties of the tissue, in particular the scattering phase and the anisotropy factor (g).

[0249] Figure 4 A functional schematic diagram of an example opto-physiological sensor system, which may be opto-electronic hardware system 400 , is shown. Figure 4The illustrated optical physiological sensor system 400 includes a sensor module 401 coupled to a microcontroller unit (MCU) 405 via an analog front end 403. The analog front end 403 includes a multiplexed LED driver, an amplifier, a bandpass filter, a bandstop filter, and a demultiplexer. It should be understood that Figures 1 to 3 The characteristics of the optical physiological sensor system can be compared with Figure 4 For example, the sensor module 401 may include Figures 1 to 3 The photodiode 101 , the temperature sensor 103 , and the plurality of illumination sources 105 , 107 , 109 , 111 and the MEMS motion sensor 135 .

[0250] The sensor module 401, the analog front end 403 and the microcontroller 405 are coupled to a power management module 411, which in turn is coupled to a battery 417. The microcontroller unit 405 is also coupled to an optional debugger 407 and an optional computer 409. A wireless communication interface 413 is coupled in parallel to the power management module 411 and the microcontroller unit 405. The wireless communication interface 413 is configured to communicate with a cloud server and a mobile application 415.

[0251] The sensor module 401 includes:

[0252] 1. Multi-wavelength illumination photoelectric sensor (called photoelectric sensor),

[0253] 2. MEMS motion sensor with 3-axis gyroscope, 3-axis accelerometer and 3-axis digital compass,

[0254] 3. Planar pressure sensor, and

[0255] 4. Four temperature sensors.

[0256] The signal captured by the sensor module 401 is divided into two parts.

[0257] 1. The first part is that the 5-channel signal captured by one or more photodiodes is initially amplified by the analog front-end electronic device (AFE) 403 and then passed to the MCU 405 after being filtered by a combination of a bandpass filter and a bandstop filter.

[0258] 2. The second part is the reference signals captured by the planar pressure sensor, four (4) temperature sensors, and MEMS sensor unit. These reference signals are directly transmitted to the MCU 405 after primary amplification.

[0259] MCU 405 plays four roles:

[0260] 1. Analog-to-digital conversion (ADC),

[0261] 2. AFE 403 control and drive,

[0262] 3. Utilize real-time signal processing embedded with artificial intelligence algorithms, and

[0263] 4. Pass and The 4G / 5G LTE wireless communication interface 413 performs real-time data / reading transmission.

[0264] Thus, a computer program and / or instructions may be included, which means that the MCU 405 is configured to receive and process sensor signals according to a photophysiological model, as described above.

[0265] These processed or unprocessed signals can be communicated via near field communication (ie, or 4G / 5G LTE) via Transmitted to mobile application 405 .

[0266] The wireless communication interface 413 is configured for over-the-air (OTA) communications by utilizing two functions:

[0267] 1. Upload the signal to the cloud server by remote communication and then send it to the mobile phone APP; and

[0268] 2. Update the control and drive systems and AI signal processing algorithms.

[0269] The power management module 411 is responsible for managing the power supply of the entire system, including all voltage conversions and battery charge and discharge management.

[0270] The debugger module 407 can be connected to a PC via USB Type-C to update the control and drive systems and AI signal processing algorithms and debug system errors.

[0271] Figure 5 A functional schematic diagram showing an example method of processing a signal via a processor such as Figures 1 to 3 Optical physiological sensor system or Figure 4 The light of the physiological sensor system is obtained from photoelectric sensors.

[0272] Raw signals are captured by a multi-wavelength illumination photosensor (photosensor or photodiode) at step 501. All raw signals are normalized from 0 to 1. The raw signals may then be pre-processed 503 (eg, via orthogonal separation, as described in more detail below).

[0273] The raw signal is then fed into the offline AI 507 and the real-time AI 505 in parallel.

[0274] Real-time AI 505 is a Generative Adversarial Network (GAN) framework consisting of two generators and one discriminator. Generator 1 (G1) aims to learn a noise-resistant mapping from an input noisy photoelectric signal to a reference signal. Generator 2 (G2) aims to learn the noise distribution captured from a 3-axis accelerometer, 3-axis gyroscope, temperature, and pressure sensors. Both generators, G1 and G2, share the discriminator (D).

[0275] The offline AI 507 is trained as follows:

[0276] G2 is trained using reference photoelectric signals captured from a 3-axis accelerometer, 3-axis gyroscope, temperature sensor, and planar pressure sensor to obtain the optimal solution via the loss function L G2 Generate a synthetic noisy photoelectric signal.

[0277] L 输入 = ‖G2(x)-p 输入 ‖1 [Equation 18]

[0278] where ‖·‖1 represents the L1 norm.

[0279] L r =1-Cor(G2(x),p 输入 ) [Equation 19]

[0280]

[0281] in

[0282]

[0283] Where λ=100,β=100.

[0284] Where x is the input of G2, which includes the reference photoelectric signal, 3-axis acceleration, 3-axis gyroscope, temperature and pressure signals, and p 输入 is the input noisy photoelectric signal.

[0285] According to the loss function L G1 , noisy photoelectric signals and synthetic noisy photoelectric signals are used to train G1.

[0286] L time_1 =||G1(x)-p ref ‖1 [Equation 21]

[0287] where ‖·‖1 represents the L1 norm.

[0288] L fre_1 =||FFT(F1(x))-FFT(p ref)||1. [Equation 22]

[0289] L G1_1 =log(D(G1(x);x))+λL time_1 +βL fre_1 [Equation 23]

[0290] Where λ=100,β=100.

[0291] L time_2 =||G1(G2(x))-p ref ||1, [Equation 24]

[0292] where ‖·‖1 represents the L1 norm.

[0293] L fre_2 =||FFT(G1(G2(x)))-FFT(p ref )||1 [Equation 25]

[0294] L G1_2 =log(D(G1(G2(x));G2(x)))+λL time_2 +βL fre_2 , [Equation 26]

[0295] Where λ=100,β=100.

[0296] Where x is the input noise PPG signal; G2(x) is the input synthetic noise PPG signal

[0297] therefore,

[0298] L G1 =0.5L G1_1 +0.5L G1_2 [Equation 27]

[0299] When all generators are fixed, by maximizing L D To train G1 and G2.

[0300] max D1 min G1,G2 L D1 =

[0301] log(D1(p ref ;x))+log(1-D1(G1(x);x))+log(D1(p ref ; G2(x)))+log(1-D1(G1(x2(x));G2(x)))[Equation 28]

[0303] Where x is the input noise photoelectric signal and G2(x) is the input synthetic noise photoelectric signal.

[0304] The process repeats the above training strategy to test the model until stability is established (validation).

[0305] Real-time Dataset: Once the multi-wavelength illumination photosensor is applied to one or more subjects for testing, a real-time data set is established. Consider a sampling rate of >1kHz per channel from the photosensor during cycling, walking, jogging and running or other regular physical strenuous activities, as per clinical monitoring standards.

[0306] Configuration settings: GANs are implemented in TensorFlow and trained on NVIDIA RTX 3080Ti or equivalent. All CNNs are trained for 200 epochs using random initialization with Adam optimizer parameters of 0.5 and 0.99, weight decay of 0, learning rate of 0.0001, batch size of 8, and no learning rate decay strategy.

[0307] Real-time AI 505 operates as follows: (Inference / Embedding on MCU)

[0308] Inference Flow: G1 can be easily ported to embedded sensor systems.

[0309] When the offline GAN 507 (training and testing) completes the training process via the cloud computing program / algorithm, the training weights are generated using the real-time data set provided by the sensor system. The training weights are shared with the reasoning process of the GAN model. Therefore, reasoning can be built into the embedded MCU 405 and used to generate the target photoelectric signal via the input noise photoelectric signal.

[0310] Thus, it should be understood that the microcontroller unit 405 is trained by obtaining a pulsatile signal providing an indication of at least one physiological characteristic of the subject, obtaining an indication of at least one physical variable as a function of time from the wearable device, transmitting the pulsatile signal to a remote computing platform (i.e., as in FIG. 4 ) via the wireless interface 413. Figure 4 The invention relates to a method for transmitting data representing a pulsation signal and an indication of a physical variable to a workstation or computing cloud shown in the figure, training a machine learning model based on the received data at a remote computing platform, and sending the coefficients / parameters from the remote computing platform to the embedded machine learning model in the wearable device after training.

[0311] It should therefore also be understood that the microcontroller unit 405 can be trained by the following operations: obtaining a pulsating signal providing an indication of at least one physiological characteristic of the object, obtaining an indication of at least one physical variable as a function of time from the wearable device, processing the pulsating signal and the indication of at least one physical variable using a machine learning model to provide an output indicative of the physiological characteristic, sending data representing the pulsating signal and the indication of the physical variable to a remote computing platform via a wireless interface, at the remote computing platform, training an improved machine learning model based on the received data, sending the improved, trained machine learning model to the wearable device, and processing the pulsating signal using the improved, trained machine learning model and processing the indication of at least one physical variable using the machine learning model to provide an output indicative of the physiological characteristic.

[0312] For strenuous activity, the signal is pre-processed 503 via orthogonal separation as described below. It should be understood that in some examples, the micro-control unit 405 is configured to determine whether the wearer / subject is performing strenuous activity, for example, based on data from the sensor module 401. For example, if the sensor module indicates that the physical variable (e.g., acceleration) is outside a selected range or above a selected threshold level, the micro-control unit 405 can determine that the wearer / subject is performing strenuous activity.

[0313] The light intensity I at the photodiode is composed of the following terms: 1) a large DC contribution I from light scattered by local tissue DC , and 2) a much smaller AC component I AC (t), including the pulsation signal (or pulsation wave or pulsation waveform) and other time-dependent effects. In case of intense activity (IA), a significant contribution to the measured AC signal can be attributed to motion artifacts. Although the contribution of motion to the overall measured signal I is small, the presence of motion can be seen in similarly small AC signals I AC In situations of high-intensity activity, motion artifacts may dominate small pulsatile signals, obscuring their characteristics and making biomarker assessment particularly difficult.

[0314] In the measurement period T M The motion captured by the 3-axis accelerometer, 3-axis gyroscope and contact force sensor during the period is the AC component I AC (t) aspects affect light-tissue interaction.

[0315] The motion factors affecting the AC signals from the 3-axis accelerometer, 3-axis gyroscope, contact force, touch skin temperature are taken into account in the raw signals captured by the photoelectric sensor system and are expressed as:

[0316] s 正交(t) = s(t) - s m (t) [Equation 29]

[0317] s 正交 (t) = s(t) - [s a (t)+s b (t)+s p (t)+s T (t)]

[0318] where s 正交 (t) is the quadrature signal with respect to time (t), s(t) is the generated signal, and s m (t) is composed of motion factors (including acceleration component s a (t), pressure component s p (t), temperature component s t (t), and the equilibrium component s b (t)).

[0319] The details of how this motion affects the light-tissue interaction, as measured by the 3-axis accelerometer and 3-axis gyroscope in the system, are revealed. AC (t) depends on the location and method of attachment to the tissue body. In any case of repeated exercise motion, a series of alternating tissue compressions and expansions are produced, which modulate the contact force detected by the contact plane pressure sensor 123. For conventional (periodic) exercise, the AC light signal is affected in two ways. First, the strain d(t) of the tissue area is driven by the periodic motion, and in the linear elastic model, each driving frequency x(t) = sin(ωt), produces a delayed response d(t) = A(ω)sin(ωt+φ(ω)).

[0320] Secondly, when compressed, the strain produces a counterforce on peripheral blood flow; increases local effective vascular resistance; reduces blood volume, thereby also reducing the amplitude of the pulsatile signal. The strain also affects tissue absorption and scattering properties through its effect on the concentration C of absorption / scattering centers.

[0321] So, for example, the parameter μ a (n) defines the absorption per unit path length of tissue type n (dermis, epidermis, etc.), which can be written in terms of the extinction coefficient ∈ and the absorber concentration C for that tissue type as follows: a (n)=∈(n)×C(n).

[0322] Consider that the detected light intensity I (once bandpass filtered) is approximated by the Taylor series

[0323]

[0324] Approximately small time correlations (oscillations ( osc )) term, I AC (t) is a linear combination of the pulsating signal term and the motion term, which consists of many delayed responses to the forcing term frequency and is equivalent to adding the forcing frequency to I AC These frequencies are then removed to recover the pulsation signal. For regular exercise, it is sufficient to include the main forcing frequencies measured by the accelerometer.

[0325] In orthogonal separation, by limiting the M 'The space inside the function' (Generally this time interval is a window of 8 to 10 seconds, advancing one second at a time) 3 -axis accelerometer reading (and its integrated velocity) a k (t), k = 1…6 and complementary orthogonal subspaces To remove these main frequencies. The orthogonal separation method obtains the pulsating signal as the measured AC signal s(t) into the orthogonal subspace projection; a construction equivalent to relative to N k The change in minimizes the integral:

[0326]

[0327] Typically, all signals must first be filtered to remove any: DC components, low frequency baseline drift, and high frequency noise, but it is critical that the bandpass filters used prior to quadrature separation are zero phase.

[0328] The processing can be done very quickly (<< 1 second), making it well suited for use in real-time physiological monitoring systems. The method is accurate when the true pulsatile signal and the accelerometer signal occupy different regions of the frequency domain. If the signals overlap in frequency space, the method is still very effective for extracting biomedical markers because the peaks in the force frequencies are significantly sharper than those in the pulsatile signal (largely due to frequency spreading caused by the presence of surrounding soft tissue).

[0329] In this case, the effect of quadrature separation is to reduce the height of the broad pulse frequency peaks, but not to remove them. Even in this rare case, there is sufficient signal integrity to obtain heart rate (HR), respiratory rate (RR), oxygen saturation (SpO2) level, pulse transit time (PTT) and pulse wave velocity (PWV) and other biomarkers.

[0330] The four wavelengths provided by the device advantageously allow all such biomarkers to be obtained at each wavelength, thereby increasing the confidence of the results and further reducing the noise through independent component analysis (ICA). The four wavelengths used in the photoelectric sensor advantageously allow additional quality control through a section of analysis that is also an important part of evaluating the oxygen saturation level (SpO2). The SpO2 level is usually obtained using the Ratio-of-Ratios method, and many authors have observed that the ratio of the AC / DC value at wavelength λ1 to the AC / DC value at wavelength λ2, σ(λ1,λ2), is a reasonable proxy that is only related to the SpO2 value.

[0331] Verification of SpO2 with gold standard data is usually well fitted to some low-degree polynomial in σ(λ1,λ2), usually linear or quadratic. The measurements of the photoelectric sensor (shown as a plot of σ(λ1:650nm,λ2:910nm)) values ​​against σ(λ1:525nm,λ2:595nm)) values) show a very good linear relationship, which means that the proxy parameters can be defined (It does not matter which pair of wavelengths is used)

[0332]

[0333] On the one hand, this supports the general assumption that the presence of a parameter is a proxy biomarker And (assuming, say, a quadratic fit for SpO2), for some a,b,c, we get Additionally, tracking the linearity of the σ-σ fit provides a useful check on signal integrity.

[0334] This is especially important in situations where the SpO2 method works worst under cold conditions such as detected by the device's temperature sensor on darker skin tones.

[0335] It will be appreciated that obtaining such biomarkers does not require a wearable device, but instead may be determined, for example, by using a camera or other device operable to receive and distinguish between different wavelengths of light.

[0336] In some examples, the microcontroller unit 405 is configured to, for example, determine the mainstream or main form of noise based on a physical variable (e.g., obtained via the sensor module 401), and apply a selected model based on the mainstream or main form of noise to remove noise in the pulsating signal. For example, if the main form of noise is due to temperature, the microcontroller unit 405 can be configured to apply a temperature-based model. In some examples, the model can be applied only when the noise reaches a selected noise threshold level, for example, if the noise from the determined source is higher than the selected threshold level. Applying the model only to the mainstream or main form of noise, rather than always applying all models, can advantageously improve the real-time processing of the signal. This in turn can advantageously improve the real-time output of the physiological parameters.

[0337] As described above, in some examples, the microcontroller unit 405 is also configured to not obtain data if the parameter is outside a selected threshold range. For example, if the temperature reading is outside a selected range, or if the contact pressure is outside a selected range, the microcontroller unit 405 is configured not to obtain any data, and / or not to send any data via the wireless communication interface 413.

[0338] As described above, the photophysiological sensors and monitoring methods can be well suited to accommodate different physiological variables, such as tissue type, age, weight, tissue location, presence of tattoos, and skin tone / complexion, with an automatic attenuation function to automatically select the lighting intensity appropriate for the operation of the photoelectric sensor.

[0339] The performance of the optical physiological monitoring system has been tested with six healthy volunteers. The system performance was evaluated for tissue type, tissue location, and skin color. The analog front end drives multiplexed LEDs (four wavelengths or more: 525nm green, 595nm orange, 650nm red, and 910nm infrared), providing a high-resolution pulsating waveform (<1ms) to capture changes in light-transmitting illuminated tissues. These wavelengths used are only exemplary wavelengths, and other wavelengths can be used, for example, in the range of 465nm to 1450nm. For example, the wavelength used can be based on individual physiological variables (such as tissue type, age, weight, tissue location, presence of tattoos, and skin tone / skin color). For example, a newborn with a light skin color can use a different wavelength than an adult with a dark skin color.

[0340] Figure 6The Fitzpatrick scale is shown (Source: Hypoxia Lab, “Protocol for skin color assessment for pulse oximeter performance studies,” August 2022, https: / / openoximetry.org / study-protocols). Photophysiological monitoring systems (using Figure 6 Fitzpatrick scale for determining skin type). Figures 7 to 12 The results shown in are for six healthy volunteers: four males (skin types I, III, VI, VI) and two females (skin types II, V). For each subject, measurements were collected at room temperature (20±3° C.) in a stationary state for one minute.

[0341] Figures 6 to 12 An optical physiological sensor (such as Figures 1 to 3 Pulsating waveforms captured by a photophysiological sensor from four wavelengths of illumination.

[0342] Figure 6 Shown is a pulsatile waveform of a subject with skin type I (Fitzpatrick scale) collected from the back of the left wrist using a photophysiological monitoring system.

[0343] Figure 7 Shown is a pulsatile waveform of a subject with skin type II (Fitzpatrick scale) collected from the back of the left wrist using a photophysiological monitoring system.

[0344] Figure 8 Shown is a pulsatile waveform of a subject with type III skin (Fitzpatrick scale) collected from the back of the left wrist using a photophysiological monitoring system.

[0345] Fig. 9 Shown is a pulsatile waveform collected from the back of the left wrist of a subject with skin type IV (Fitzpatrick scale) using a photophysiological monitoring system.

[0346] Fig.10 Shown is a pulsatile waveform of a subject with type V skin (Fitzpatrick scale) collected from the back of the left wrist using a photophysiological monitoring system.

[0347] Fig.11 Shown is a pulsatile waveform of a subject with skin type VI (Fitzpatrick scale) collected from the back of the left wrist using a photophysiological monitoring system.

[0348] Fig.12 Shown is a pulsatile waveform of a subject with type V skin (Fitzpatrick scale) collected from the back of the left wrist using a photophysiological monitoring system.

[0349] With Masimo TFA-1 TM Compared with the disposable transflective sensor and the NONIN 8000R reflective SpO2 sensor, as mentioned above, Figure 1 The SpO2 comparison results of the optical physiological monitoring system are shown in Table 1 below. The measurement site for this test was the forehead. The skin type was determined using the Fitzpatrick scale. Healthy volunteers with skin types II, III, V, and VI were tested. For each subject, measurements were collected at room temperature (20 ± 3 ° C) in a static state for one minute.

[0350] Table 1 Photophysiological monitoring system, Masimo TFA-1 TM Comparison of SpO2 results between disposable transflective sensor and NONIN 8000R reflective SpO2 sensor

[0351]

[0352] As mentioned above about Figure 1 The photophysiological monitoring system is combined with the NONIN 8000R reflectivity SpO2 sensor and Masimo TFA-1 TM The results of the perfusion index (PI) comparison compared to the disposable transflective sensor are shown in Table 2. The measurement sites for this test were the wrist, index finger, and forehead. The skin type was determined using the Fitzpatrick scale. Healthy volunteers with skin types II, III, V, and VI were tested. For each subject, measurements were collected for one minute at rest at room temperature (20 ± 3 °C). Figure 8 The results shown in are for healthy volunteers with skin type VI.

[0353] Fig.13 Pulsation waveforms of male subject type VI skin (Fitzpatrick scale) monitored with 4 wavelengths of light physiological monitoring. Subject in resting state, measurement site: back of left wrist. Room temperature: 20°C (±2).

[0354] Table 2. PI (%) average results for the optical physiological sensor of the present disclosure for four subjects (skin types II, III, V, VI) and two commercial clinical-grade pulse oximetry probes.

[0355]

[0356] The embodiments shown in the drawings are exemplary only and include features that may be generalized, removed or replaced as described herein and as set forth in the claims. Other examples and variations of the apparatus and methods described herein will be apparent to those skilled in the art in the context of the present disclosure.

Claims

1. A method for monitoring an object using a photophysiological sensor system, the photophysiological sensor system comprising at least one sensor, the method comprising: obtaining a model of optical properties of at least one body tissue type to be monitored, wherein said model of said optical properties comprises a definition of a static component and a dynamic component of a transmitted optical power and a definition of a source-detector spacing related to a normalized path length of an illumination source for said optical physiological sensor; obtaining an indication of at least one physiological characteristic of the subject from a wearable device including the at least one sensor worn by the subject; obtaining an indication of at least one physical variable from the wearable device including the at least one sensor worn by the subject; determining, using the model, how the at least one physical variable affects the at least one physiological characteristic; and Based on the determination of how the at least one physical variable affects the at least one physiological characteristic, a corrected value for the physiological characteristic is determined.

2. The method according to claim 1 further comprises using the corrected value for the physiological characteristic to determine a measurement value of an accurate physiological parameter, the physiological parameter comprising at least one of the following: heart rate HR, perfusion index PI, oxygen saturation SpO2%, respiratory rate RR, blood pressure BP, pulse rate variability PRV, pulse transit time PTT and pulse wave velocity PWV.

3. The method for monitoring an object using an optical physiological sensor according to claim 1, wherein: The at least one physical variable includes at least one of: contact pressure, contact temperature, acceleration, angular velocity, and absolute orientation.

4. A method according to any one of the preceding claims, wherein: The model includes defining the minimum optical power dP penetrating into a point (x, y) on the surface of the body tissue type according to equation 1: ρ(λ, l′(x′, y′)): dP(λ, x, y) = ∫∫I0(λ, x′, y′) × e -ρ(λ,l′(x′,y′)) dx′dy′ [Equation 1] wherein I0(λ,x,y) represents the total light entering through the entire surface of the body tissue type, and I0(λ,x,y) undergoes an exponential decay that is a function of the optical density ρ(λ) of the body tissue type; Wherein, the source-detector spacing is based on the s ,y s ) and an arbitrary detector at a point (x, y) on the surface of the body tissue type, such that and Wherein, for the surface of the body tissue type, the vector x and the vector y from x - Cross to x + and from y - Cross to y + For a detector with a finite rectangular area defined, the optical power received by the detector is according to Equation 2:

5. A method according to any one of the preceding claims, wherein: The model includes the use of the normalized physiological pulse function ψ(t) according to Equation 3 and the respective absorption coefficient μ ai (λ), scattering coefficient μ si (λ) and pulse coefficient μ pi (λ), defines the optical response of a dynamic and multi-layered body tissue type in terms of its dynamic optical density ρ(λ,l′,t), where the number of layers i ranges from 1 to N:

6. The method according to claim 4, wherein: The model involves separating the static component of Equation 4 according to Equations 5 to 6 and dynamic component 7. The method according to claim 6, wherein: The model includes defining the static component of the transmitted optical power for a rectangular detector defined by vectors x and y using the integral-sum rule for Equations 1 and 2 according to Equations 7, 8, and 9: and dynamic component 8. The method according to claim 7, wherein: The model includes defining the optimal source-detector spacing l'(λ) for an illumination source at wavelength λ according to Equation 10:

9. A method according to any one of the preceding claims, wherein: The model includes the assumption of cylindrical symmetry and optical homogeneity of the body tissue type to be monitored, so that the optimal source-detector spacing l'(λ) is expressed as the position of the illumination source (x s ,y s ) as the center, or conversely, the position of the detector (x d ,y d ) as the center:

10. A method according to any one of the preceding claims, wherein: The at least one physical variable includes at least one of acceleration, angular velocity, and absolute orientation, and wherein using the model to determine how the at least one physical variable affects the at least one physiological characteristic includes modeling a pumping action of the heart to determine a volumetric blood flow rate.

11. The method according to claim 10, wherein: Modeling the pumping action of the heart includes determining the pressure gradient in the form of Equation 12: where A0 is the constant component of the blood pressure gradient from the heart, A1 is the amplitude of the fluctuating component, and ω p =2πf p , where f p is the pulse frequency, and the volumetric blood flow rate Q is determined by Equation 13 and Equation 14: Among them, ρ and μ f are the density and viscosity of the blood flowing through the blood vessels, and u z is the velocity of blood flow in the axial direction, and wherein the z-axis is taken along the axis of the arterial blood segment, and r is taken radially as a combination of acceleration, angular velocity, and absolute orientation that can be obtained when obtaining an indication of at least one physical variable from the wearable device.

12. A method according to any one of the preceding claims, wherein: The at least one physical variable comprises temperature, and wherein determining how the at least one physical variable affects the at least one physiological characteristic using the model comprises modeling density of a particular tissue type as a function of wavelength and temperature of the illumination source of the optical physiological sensor.

13. The method according to claim 12, wherein: Modeling the density of a particular tissue type as a function of the wavelength and temperature of the illumination source of the optical physiological sensor includes modeling tissue temperature as a result of blood perfusion defined by thermoregulation according to Equation 15, which reflects changes from tissue composition, skin thickness, surface area, tissue volume, and ambient temperature in the presence of living tissue properties or body tissue environment: where ρ(λ,t) indicates the density of a specific tissue type, C p is the tissue specific heat, V is the specific tissue volume, and A is the tissue surface area, such that A = πD 2 / 2, and wherein V and A depend on the average tissue diameter D=2R, where R is the average radius of the tissue surface area A; and wherein the density and specific heat of blood are respectively given by ρ b and C pb Indicates that, while ω b is the volumetric blood flow, and where T A is the arterial temperature, h 空气 is the heat transfer coefficient at the skin surface dominated by ambient heat exchange, and T 空气 Indicates the temperature of the environment.

14. The method according to claim 13, wherein: Modeling the density of a specific tissue type as a function of the wavelength and temperature of the illumination source of the optical physiological sensor includes modeling tissue temperature as a result of blood perfusion defined by thermoregulation of multiple layers using multi-wavelength illumination according to Equation 16: Among them, k N , N , C pN ,ω bN , and q mN represents the thermal conductivity, density, specific heat, blood perfusion and metabolic heat production of the corresponding tissue layer (N), ρ b is the blood density, and represents the specific heat of blood, and T A is body temperature and is considered a constant.

15. A method according to any one of the preceding claims, wherein: The at least one physical variable comprises a contact pressure of the optical physiological sensor with the subject's skin, and wherein determining how the at least one physical variable affects the at least one physiological characteristic using the model comprises modeling changes in optical density of a specific tissue type based on changes in the optical path length.

16. The method according to claim 15, wherein: Using the model to determine how the at least one physical variable affects the at least one physiological characteristic includes: when the contact force is within a selected range, modeling the optical density of the specific tissue type as a function of temperature and wavelength as proportional to the contact force; and when the contact force is outside the selected range, modeling the optical density of the specific tissue type as a function of temperature and wavelength as proportional to the inverse of the contact force.

17. The method according to claim 15 or 16, comprising modeling the change in optical path length due to an increase or decrease in contact force as and The dynamic AC component and the static DC component of the light density give rise to a change in the light density ρ(λ,t) such that 18. A wearable optical physiological sensor system, the wearable optical physiological sensor system is used to obtain physiological characteristics of a wearer and physical variables of the wearer's activities, the optical physiological sensor system comprising: A photosensor panel comprising a plurality of photodiodes and a plurality of illumination sources, wherein at least one of the plurality of illumination sources is selectively configured to emit light of a different wavelength than the other illumination sources, such as light emitting diodes; a planar pressure sensor for obtaining an indication of contact pressure between the optical physiological sensor and the wearer; and A printed circuit board includes a micro-electromechanical sensor for obtaining the physical variable, a micro-control processor configured to process signals received from the plurality of micro-electromechanical sensors and the plurality of photodiodes, and a wireless communication interface.

19. The wearable optical physiological sensor system according to claim 18, wherein: The micro-electromechanical sensor includes at least two of a gyroscope, an accelerometer, and a digital compass.

20. The wearable optical physiological sensor system according to claim 18 or 19, wherein: The photoelectric sensor panel is separated from the planar pressure sensor by a flexible substrate.

21. A wearable optical physiological sensor system according to claim 18, 19 or 20, wherein: The planar pressure sensor is separated from the printed circuit board by a flexible substrate.

22. The wearable optical physiological sensor system according to any one of claims 18 to 21, wherein: The printed circuit board is flexible.

23. The wearable optical physiological sensor system according to any one of claims 18 to 22, wherein: The photosensor panel includes a plurality of temperature sensors configured to obtain an indication of the wearer's skin by contacting the wearer's skin.

24. The wearable optical physiological sensor system of claim 23, wherein: The photosensor panel includes a plurality of temperature sensors equally spaced apart on the photosensor panel.

25. The wearable optical physiological sensor system according to any one of claims 18 to 24, wherein: The processor is configured to transmit raw sensor data or real-time processed data via the wireless communication interface.

26. The wearable optical physiological sensor system according to any one of claims 18 to 25, wherein: The processor is configured to apply a machine learning model to sensor data obtained from the plurality of photodiodes, pressure data obtained from the planar pressure sensor, and the physical variable to generate modified output data, and to transmit the modified output data via the wireless communication interface.

27. A wearable optical physiological sensor system according to any one of claims 18 to 26, wherein: The processor is configured to not obtain sensor data from the plurality of photodiodes or micro-electromechanical sensors if one or more temperature sensors provide an indication of a temperature outside of a selected temperature range.

28. The wearable optical physiological sensor system according to any one of claims 18 to 27, wherein: The microcontrol processor is configured to not obtain sensor data from the plurality of photodiodes or the micro-electromechanical sensor if the planar pressure sensor provides an indication of a contact pressure outside of a determined pressure range.

29. A method for monitoring a subject using an optical physiological sensor system, the method comprising: obtaining a signal from a wearable device worn by the subject, the signal comprising a component indicative of at least one physiological characteristic of the subject and a component indicative of at least one physical parameter; modeling the obtained signal as a Taylor series based on the indication of the at least one physical parameter, the Taylor series being a linear combination of the pulsatile signal and the motion term; obtaining a corrected signal as a projection of the modeled pulsatile signal onto an orthogonal subspace; as well as At least one physiological characteristic of the subject is determined using the corrected signal.

30. The method of claim 29, wherein: Obtaining the corrected signal as the projection of the modeled signal on the orthogonal subspace comprises: M minimize the integral shown in Equation 30, where s(t) is the acquired signal, N k represents an undetermined multiplier, and a k represents three values ​​indicating motion and its integrated velocity, and the orthogonal subspace is related to a k The subspace of is orthogonal to the subspace of:

31. The method according to claim 30, wherein: The integration is performed over a rolling time window.

32. The method according to any one of claims 29 to 31, wherein: The signal comprises a dynamic AC component and a static DC component, and wherein obtaining the corrected signal comprises obtaining a projection of the measured AC signal onto an orthogonal subspace.

33. The method according to any one of claims 26 to 29, further comprising: The signal is filtered prior to minimizing the integral to remove any DC component, low frequency baseline offset and / or high frequency noise.

34. The method according to any one of claims 29 to 33, further comprising: The signal is filtered with a zero-phase bandpass filter before obtaining the corrected signal.

35. The method according to any one of claims 29 to 34, wherein: Obtaining a signal from the wearable device worn by the subject includes obtaining a plurality of pulsation signals from the wearable device, wherein each of the plurality of pulsation signals is an optical signal at a different wavelength.

36. The method of claim 35, further comprising: Independent component analysis or principal component analysis is performed on each of the plurality of pulsatile signals.

37. A method according to any one of claims 29 to 36, wherein: Obtaining an indication of at least one physical parameter from the wearable device worn by the subject includes receiving a plurality of signals from the wearable device indicative of the respective physical parameters, and normalizing the received plurality of signals between 0 and 1 to avoid any non-compliance with physical units.

38. A method for monitoring a subject using an optical physiological sensor system, the method comprising: obtaining a pulsatile signal from an optical sensor, the pulsatile signal providing an indication of at least one physiological characteristic of the subject, the indication being from a wearable device worn by the subject; obtaining an indication of at least one physical variable from the wearable device worn by the subject; determining a dominant source of noise in said pulsation signal based on said obtained indication of said at least one physical variable; determining a selected noise removal model from a list of selected models to be applied to apply to the pulsation signal; as well as A corrected value for the physiological characteristic is determined using the selected noise removal model.

39. The method of claim 38, wherein: The list of selected models includes: Orthogonal separation; Acceleration- and / or balance-based models; Temperature-based models; and Model based on contact pressure.

40. The method according to claim 38 or 39, wherein: Determining a dominant noise source in the pulsation signal based on the obtained indication of the at least one physical variable comprises determining whether the obtained indication of the at least one physical variable exceeds a selected threshold, and wherein Determining a selected noise removal model from a selected list of models to be applied to apply to the pulsation signal includes selecting the noise removal model based on a determination of whether the obtained indication of the at least one physical variable exceeds a selected threshold.

41. The method according to any one of claims 38 to 40, further comprising: obtaining a model of optical properties of at least one body tissue type to be monitored, wherein said model of said optical properties comprises a definition of a static component and a dynamic component of a transmitted optical power and a definition of a source-detector spacing related to a normalized path length of an illumination source for said photophysiological sensor system; and Wherein, using the selected noise removal model to determine the corrected value for the physiological property comprises using the model of the optical properties of the at least one body tissue type and the selected noise removal model to determine the corrected value for the physiological property.

42. The method according to any one of claims 38 to 41, wherein: The physical variables include at least one of the following: contact pressure F c , Temperature T A , acceleration a x,y,z , angular velocity Ω x,y,z , and absolute orientation ° x,y,z .

43. A method of training a machine learning model for monitoring a subject using a wearable device including a photophysiological sensor system, the method comprising: obtaining a pulsatile signal, the pulsatile signal providing an indication of at least one physiological characteristic of the subject; obtaining, from the wearable device, an indication of at least one physical variable as a function of time; transmitting data representing said pulsation signal and said indication of said physical variable to a remote computing platform via a wireless interface; training a machine learning model based on the received data at the remote computing platform; as well as After the training, the coefficients / parameters are sent from the remote computing platform to the embedded machine learning model in the wearable device.

44. The method of claim 43, wherein: Training a machine learning model based on the received data at the remote computing platform includes training a generative adversarial network (GAN), wherein the GAN includes two generators and one discriminator.

45. The method of claim 44, wherein: One of the two generators is configured to learn a noise-resistant mapping from the input noise pulsation signal to a reference signal, and the other of the two generators is configured to learn a noise distribution in the physical variable captured from the wearable device.

46. ​​A method of training a machine learning model for monitoring a subject using a wearable device including a photophysiological sensor system, the method comprising: obtaining a pulsatile signal, the pulsatile signal providing an indication of at least one physiological characteristic of the subject; obtaining, from the wearable device, an indication of at least one physical variable as a function of time; processing the pulsatile signal and the indication of the at least one physical variable using a machine learning model to provide an output indicative of the physiological characteristic; transmitting data representing said pulsation signal and said indication of said physical variable to a remote computing platform via a wireless interface; At the remote computing platform, training an improved machine learning model based on the received data; sending the improved, trained machine learning model to the wearable device; as well as The pulsatile signal is processed using the improved, trained machine learning model and the indication of the at least one physical variable is processed using the machine learning model to provide an output indicative of the physiological characteristic.

47. A method according to any one of claims 43 to 46, wherein: While a user of the wearable device is performing a selected exercise routine, obtaining a pulsatile signal from the wearable device providing an indication of at least one physiological characteristic of the subject and obtaining an indication of at least one physical variable as a function of time are performed, and wherein the machine learning model is trained at the remote computing platform based on the exercise routine.

48. A wearable optical physiological sensor system, the wearable optical physiological sensor system is used to obtain physiological characteristics of a wearer and physical variables of the wearer's activities, the optical physiological sensor comprising: a photosensor panel comprising a plurality of photodiodes and a plurality of illumination sources, wherein at least one of the plurality of illumination sources is selectively configured to emit light of a different wavelength than the other illumination sources; and a printed circuit board comprising a micro-electromechanical sensor for obtaining the physical variable, a processor configured to process signals received from the plurality of micro-electromechanical sensors and the plurality of photodiodes, and at least one wireless communication interface; Wherein, the plurality of illumination sources are located at intersections between a plurality of concentric circles, wherein each of the plurality of concentric circles is centered on a corresponding one of the plurality of photodiodes.

49. The wearable optical physiological sensor system of claim 45, wherein: The plurality of illumination sources includes at least four different illumination sources, such as light emitting diodes, each illumination source being configured to emit light of a different wavelength.

50. A wearable optical physiological sensor system according to claim 48 or 49, wherein: The plurality of illumination sources include at least sixteen different illumination sources, each illumination source being configured to emit light of one of four or more different wavelengths, wherein the at least sixteen different illumination sources are grouped into four groups, each group being configured to emit light of one of four or more different wavelengths.

51. A wearable optical physiological sensor system according to claim 49 or 50, wherein: The shortest wavelength illumination source is located at the intersection of the inner circles of the plurality of concentric circles, and wherein the longer wavelength illumination source is located at the intersection of the concentric circles further outward from the center of the concentric circles, such that the longer wavelength illumination source is positioned farther away from each of the photodiodes, and the wavelength of the illumination source increases with radial distance from each photodiode.

52. A wearable optical physiological sensor system according to any one of claims 48 to 51, wherein: The photosensor panel includes one photodiode, two photodiodes and three photodiodes.

53. A wearable optical physiological sensor system according to any of claims 49 to 50 or appended thereto, wherein: At least one of the wavelengths is about 465 nm or about 1450 nm.

54. A wearable opto-physiological sensor system according to any one of claims 48 to 53, further comprising a planar pressure sensor for obtaining an indication of contact pressure between the opto-physiological sensor and the wearer.

55. A method of obtaining an oxygen saturation biomarker of a subject, the method comprising: The object is illuminated with at least four different wavelengths λ1, λ2, λ3, λ4 to obtain a pulsating signal at each of the four different wavelengths λ1, λ2, λ3, λ4, wherein the wavelengths are selected to satisfy the condition for certain A, B, C, D: ρ=Aσ(λ1,λ2)+B=Cσ(λ3,λ4)+D [Equation 32] where σ(λ1,λ2)=r(λ1) / r(λ2); a condition well met by current devices for the chosen wavelength pairing and implied by the most current oxygen saturation SpO2%; For four different wavelengths, the ratios r(λ) of the height of the pulsatile signal to the associated DC background level are obtained and these values ​​are used in pairs to obtain the oxygen saturation biomarker of the subject.

56. The method of claim 55, wherein: The value of p obtained from Equation 19 enables fitting of the oxygen saturation biomarker to a generic, wavelength-independent curve according to Equation 33, SpO2 = SpO2(ρ) = α + βρ + γρ 2 …, [Equation 33].

57. A computer-readable non-transitory storage medium comprising a program for a computer, the program being configured to cause a processor to perform the method according to any one of claims 1 to 17, 29 to 47, and 55 and 56.

Images