RU2805810C1 - Wearable device with function of determining hemoglobin concentration, method and system for determining hemoglobin concentration - Google Patents

Abstract

FIELD: medicine.

SUBSTANCE: group of inventions relates to a method of determining the concentration of hemoglobin in the user’s blood, a wearable device and a system for determining the concentration of hemoglobin in the user’s blood. The device contains a photoplethysmographic PPG sensor and is designed to perform the method. When performing the method, a photoplethysmographic sensor is used to irradiate the user's biological tissue with radiation of two different wavelengths and detect photoplethysmographic signals in the time domain in reflection mode. The detected PPG signals are converted from the time domain to the frequency domain or time-frequency domain. The most significant signal, characterized by the highest signal-to-noise ratio, the highest amplitude and/or the least noise, is selected from all PPG signals and used as a reference signal. Harmonics are extracted from the reference signal and used as coordinates. All PPG signals are laid out along the mentioned coordinates to obtain a set of amplitude-phase characteristics of the harmonics. The concentration of hemoglobin in the user's blood is determined by comparing the resulting set of amplitude-phase characteristics of harmonics with sets of amplitude-phase characteristics of harmonics from the database, each of which corresponds to a specific value of hemoglobin concentration. The system comprises a wearable device and a remote server and/or cloud storage, a communication channel for transmitting data between the wearable device and the remote server and/or cloud storage.

EFFECT: method, wearable device, and a system for continuously monitoring hemoglobin concentration resistant to motion artifacts is provided.

32 cl, 16 dwg

Description

Field of technology

The present invention relates to devices and systems for non-invasive, personal and/or on-demand monitoring of human health, in particular, the determination of total hemoglobin (Hb) concentration, as well as its various forms, such as oxyhemoglobin, deoxyhemoglobin, methemoglobin, carboxyhemoglobin, and other blood properties, such as oxygenation. In addition, the invention relates to a method for determining hemoglobin concentration, implemented in wearable devices, in particular, in modern smart watches and fitness bracelets.

BACKGROUND OF THE INVENTION

Hemoglobin is the main respiratory pigment and the main component of red blood cells, which performs important functions in the human and animal body: transporting inhaled oxygen from the lungs to biological tissues and organs, and carbon dioxide from tissues and organs to the lungs, where it is exhaled. This occurs due to the fact that oxygen is able to reversibly bind with iron, the atoms of which are “embedded” in hemoglobin. When interacting with biological tissues and organs, red blood cells are “freed” from oxygen and “take up” carbon dioxide. Hemoglobin also plays a significant role in maintaining the acid-base balance of the blood. The buffer system created by hemoglobin helps maintain blood pH within certain limits.

One red blood cell contains about 3.4 x 10 ⁸ hemoglobin molecules, each of which consists of approximately 103 atoms. On average, human blood contains about 14.5% hemoglobin, its total amount is about 750 g. Hemoglobin is a complex protein belonging to the group of hemoproteins, the protein component of which is represented by globin, the non-protein prosthetic group. The prosthetic group in the hemoglobin molecule is represented by 4 identical iron porphyrin compounds, which are called themes. The heme molecule consists of porphyrin IX linked to iron by two nitrogen atoms by covalent bonds and two other nitrogen atoms by coordination bonds. The iron (II) atom is located in the center of the heme and gives the blood its characteristic red color; its oxidation state does not change regardless of the addition or release of oxygen.

As discussed above, the most characteristic property of hemoglobin is the reversible addition of gases such as O ₂ , CO ₂ , etc. The resulting compounds are called oxyhemoglobin and carboxyhemoglobin, respectively. The reaction of addition of molecular oxygen is not a true oxidation of hemoglobin, since the valence of iron in the substance does not change, and this reaction is more correctly called oxygenation. True oxidation of hemoglobin occurs only when iron passes into the ferric state.

In the blood, hemoglobin exists in at least four forms: oxyhemoglobin, deoxyhemoglobin, carboxyhemoglobin, and methemoglobin. In erythrocytes, the molecular forms of hemoglobin are capable of interconversion; their ratio is determined by the individual characteristics of the organism.

Normal values for hemoglobin concentration: in men 130-165 g/l, in women 120-150 g/l, in children 110-155 g/l, in pregnant women 110-120 g/l. A diagnostically significant decrease in the lower threshold of normal hemoglobin levels occurs in men in the age group 65-74 years.

The concentration (level) of hemoglobin increases when the number of red blood cells increases and decreases when there are fewer red blood cells. A decrease in the number of red blood cells, and therefore a reduced concentration of hemoglobin in the blood (low hemoglobin), may be due to a decrease in the formation of red blood cells in the bone marrow, their loss as a result of bleeding or destruction within the body, unbalanced nutrition, bad habits, impaired hemoglobin synthesis, gastrointestinal diseases tract, chronic kidney disease, liver cirrhosis, myxedema, hemolysis, etc. The consequences of low hemoglobin are disruptions of the immune system, deterioration of the skin, a negative impact on the ability to motherhood, tachycardia, shortness of breath, provoking a stroke and heart attack. In this case, a person is often in a state of anemia, when the body does not receive enough oxygen, which causes weakness, drowsiness and fatigue.

An increased concentration of hemoglobin indicates polycythemia, hemoconcentration due to dehydration, burns, intestinal obstruction, persistent vomiting, exposure to high altitudes, and excessive physical activity; cardiovascular pathology, usually congenital, leading to significant venous discharge; lung diseases leading to decreased pulmonary perfusion, poor lung aeration, pulmonary arterial fistula; chronic chemical exposure to nitrites, sulfonamides, causing the formation of meth- and sulfohemoglobin.

Related to hemoglobin concentration is blood oxygen saturation (oxygenation or SpO ₂ ), which is the percentage of oxyhemoglobin in the blood, that is, the ratio of the amount of oxyhemoglobin to the total amount of hemoglobin. For example, oxygenation of 95-100% is normal, but may vary if a person has lung disease. Oxygenation <90% is considered low and is called hypoxemia, and the body requires oxygen supplementation. The causes of low blood oxygen saturation are anemia, sleep apnea, smoking, lung diseases (asthma, emphysema, etc.), viral infections (COVID-19, etc.), air quality, reduced cardiac capacity, strong painkillers. The consequences of low oxygenation include headaches, tachycardia, shortness of breath, and damage to the heart and brain. It should be noted that the level of oxygenation reflects the degree of blood saturation with oxyhemoglobin, but does not characterize the total hemoglobin content in the blood.

Thus, hemoglobin concentration and oxygen saturation of human blood are among the most important and significant indicators of human health and require constant monitoring.

To determine hemoglobin, invasive and non-invasive methods are used. Invasive methods are laboratory methods, which include chemical, spectrophotometric, colorimetric methods, such as the saponin method, Sali method, etc., when hemoglobin derivatives formed during its oxidation and addition to various chemical groups are most often analyzed, leading to a change in the valence of iron and the color of the solution. Chemical and spectrophotometric methods are highly accurate and are recommended as reference methods, but due to the laboriousness and significant cost of analysis, they are not used for routine determinations. For routine laboratory tests, colorimetric methods are most preferred, as they are the cheapest, simplest and fastest to perform. The ABG test (arterial blood gas test) is used to determine blood oxygen saturation.

However, methods for monitoring hemoglobin and blood oxygen saturation associated with blood sampling and testing are not suitable for continuous monitoring and are inconvenient. The advantages of measuring hemoglobin concentration using wearable devices, in particular smart watches or fitness bracelets, are a non-invasive measurement method, fast measurement, ease of use, and the possibility of continuous and long-term monitoring of hemoglobin. In addition, a rapid oxygen saturation test can be useful for ongoing monitoring and decision-making regarding hospitalization of the patient.

Non-invasive methods include, for example, pulse oximetry, which monitors the percentage of hemoglobin saturated with oxygen (O ₂ ) by assessing tissue transmission of optical radiation (based on pulse amplitude) and heart rate. To non-invasively determine blood oxygenation, a tissue section containing arterial vessels is placed in the working area of the photoplethysmographic (PPG) sensor. Traditional oximeters usually have the form of a “clothespin” to be placed on the patient’s finger or earlobe, that is, where radiation can be transmitted through a tissue sample, and analyze the radiation passing through the finger or earlobe, that is, they measure the tissue sample for transmission. In human tissues, bones, venous blood and arterial blood, radiation is absorbed, reflected and scattered, which, when studying blood flow, is determined by the size of the vessels or the volume of blood passing through the tissue area being examined. The narrowing and expansion of the vessel under the influence of arterial pulsation of blood flow, which is caused primarily by changes in blood flow in the arteries and arterioles, cause a corresponding change in the amplitude of the signal received from the output of the photodetector. The pulse oximetry technique is based on the principles of photoplethysmography (PPG). Hemoglobin serves as a kind of filter for light flux, and its “color” and “thickness” can change. The “color” of the filter depends on the percentage of oxyhemoglobin. This is the basis for the ability of a pulse oximeter to determine the degree of blood oxygenation. The change in the “thickness” of the filter is affected by the increase in blood volume in the arteries and arterioles with each pulse wave. After measuring the pulse rate and pulse wave amplitude, the ratio of the degree of absorption of radiation of different wavelengths, for example, IR and red waves, is analyzed using a microprocessor, then the saturation of the pulsating flow of arterial blood with oxygen is calculated.

In this case, the signal from the output of the sensor, proportional to the absorption of light passing through the tissue, includes two components: a variable (AC) component, caused by a change in the volume of arterial blood with each heartbeat, and a constant (DC) component, determined by the optical properties of the skin, bones, venous blood and other tissues of the area under study, the total absorption of which does not change during the propagation of the pulse wave.

This is a traditional approach for signal analysis in pulse oximetry, based on the separation of the constant and variable components of the time series (DC and AC components in the time domain or direct and alternating current components) of the PPG signal (pulse signal). In this case, the integral area of the signal curve is traditionally measured, and the signal shape is usually not taken into account. At the same time, the DC component is used to normalize the signal, since signal transmission depends not only on the absorption/reflection/scattering coefficients of skin, bones, blood and other tissues that make up the “filter” for the signal, but also on the thickness of the “filter” itself.

Most known methods describe the theory and operation of pulse oximeters in transmission mode, therefore, in the prior art, such approaches are described based on the relevant laws and equations, in particular, the Bouguer-Lambert-Beer law, which determines the attenuation of a parallel monochromatic light beam as it propagates in an absorbing medium . However, there is not enough information in the prior art regarding the application of the above method with AC/DC components in a pulse oximeter in reflection mode. It is obvious that operation of the device in reflection mode requires other approaches, including additional analysis of the pulse waveform.

However, there are certain difficulties in analyzing the AC components of the PPG signal, consisting in the fact that the AC component has a complex shape, which depends on the wavelength and carries information about the spectral properties of the tissue.

Existing devices for determining hemoglobin have low accuracy and are sensitive to motion artifacts, and therefore are not suitable for use as consumer gadgets. In addition, for example, pulse oximeters have an additional device - a clothespin on the finger and are not suitable for continuous monitoring of blood oxygen saturation.

US Pat. No. 8,255,028 B2 (Masimo Corporation) discloses a patient monitor that has multiple sensors capable of attaching to tissue sites in a living subject. The sensors generate sensory signals that respond to at least two wavelengths of optical radiation after attenuation by pulsating blood in tissue areas. The patient monitor uses multiple signals to reduce the impact of noise.

The disadvantages of this solution are the non-use of information about the shape of the pulse wave and the lack of correction for the scattering/absorbing properties of the tissue.

US Pat. No. 9,341,565 B2 (Masimo Corporation) discloses a physiological monitor for determining blood oxygen saturation in a patient being treated, including a sensor, a signal processing device, and a display device. The sensor includes at least three light-emitting diodes. Each light-emitting diode is capable of emitting light of a different wavelength. The sensor also includes a detector capable of receiving light from at least three light-emitting diodes after being attenuated by tissue. The detector generates an output signal based at least in part on the received light. The signal processing device determines blood oxygen saturation based on at least the output signal, and the display device provides an indicator of blood oxygen saturation.

The disadvantages of this solution are also the non-use of information about the shape of the pulse wave and the lack of correction for the scattering/absorbing properties of the tissue.

Below is Table 1 showing the comparative features of the proposed invention and other solutions known in the prior art.

Solutions of the prior art, as a rule, provide for the operation of sensors in transmission mode and do not fully disclose the principles of operation of wearable devices with the function of determining blood properties, such as hemoglobin concentration, in reflection mode, which is used, for example, in the form factor of smart watches.

For example, recent publications such as WO 2023/003980 A1 and WO 2023/287789 A1 (Masimo Corporation) disclose wearable devices in the form of smart watches for monitoring physiological parameters of a user, which can be placed on the wrist, with arrays of radiation sources and detectors located on one side of the devices. At the same time, when describing the physical principles of operation of the PPG sensor, the authors of these publications refer to the Bouguer-Lambert-Beer law, as noted above in relation to other solutions of the prior art. However, when working in reflection mode, this approach does not work and gives incorrect results. In addition, signal processing is performed using traditional pulse oximetry signal processing methods, i.e. These publications focus on the design of the sensor, without describing specific signal processing methods, especially in relation to the measurement of specific blood components, such as hemoglobin. In standard mode, when the sensor is in the watch on the wrist, it is usually possible to measure heart rate, heart rate variability and oxygenation. To measure specific parameters, an external sensor can be used, connected to the watch port and placed, for example, on the user’s finger. Thus, these publications do not disclose the features of processing signals received in reflection mode, which make it possible to measure specific blood parameters, such as hemoglobin concentration. Those. these publications do not contain information about the shapes of the PPG signal at different wavelengths, the differences in the optical paths of radiation at these wavelengths, the differences in the types of vessels and tissues with which radiation of different wavelengths predominantly interacts and, accordingly, about the way to take into account the differences in optical paths and types of vessels and tissues through the shape of the PPG signal and the method of describing the shape of this signal through the amplitude-phase characteristics of harmonics.

Thus, the main problem in this field of technology is the lack of a method and easy-to-use wearable device for continuous monitoring of blood properties, for example, hemoglobin concentration, especially in the mode of operation of sensors in reflection mode, as well as the instability of existing solutions to movement artifacts.

Therefore, there is a need for a wearable device and method for determining blood properties, in particular hemoglobin concentration, that provides increased spectral resolution and is suitable for non-professional use, i.e. a reliable and motion-resistant solution is needed. In other words, a device is required for non-invasive, unobtrusive, continuous monitoring of hemoglobin concentration and blood oxygen saturation.

Possible products that use the method of the invention are wearable devices such as smart watches or smart bracelets, stationary diagnostic devices, household appliances and personal health monitoring gadgets.

The essence of the invention

In view of the above technical problems, the present invention will now be described by way of example and not by way of limitation with reference to the description and drawings below.

This summary of the invention precedes a detailed description of specific exemplary embodiments.

implementation to give a general overview of aspects of the claimed invention that will be further explained below, and is in no way intended to define or limit the scope of the present invention.

The objective of the present invention is to create wearable devices with the function of determining the concentration of hemoglobin in the user's blood based on data collected from the sensors of wearable devices.

The proposed wearable device, method and system provide the possibility of personal, non-professional use, resistance to movement and the ability to determine blood properties, in particular, hemoglobin concentration with high accuracy.

A wearable device and system are proposed that implement a method for determining the hemoglobin concentration in the user's blood, including the steps of: irradiating the user's tissue with a light source emitting light of at least two different wavelengths, detecting photoplethysmographic (PPG or pulse) signals from the user's tissue at the mentioned at least two different wavelengths, convert the detected PPG signals from the time domain to the frequency domain, select, from all the PPG signals available at several wavelengths, the most significant signal with the best signal-to-noise ratio, the highest amplitude and the least affected noise and use it as a reference signal, extract frequency components (harmonics) from the reference signal and use them as coordinates, lay out the rest of the signals according to the coordinates obtained from the reference signal, obtaining a set of characteristics, and compare the set of characteristics with sets from a pre-compiled database, determining hemoglobin concentration without a laboratory blood test.

Brief description of drawings

The above and other features and advantages of the present invention will be explained in the following description with reference to the accompanying drawings. The following specific exemplary embodiments of the claimed invention, taken in conjunction with the drawings, are not intended to limit the scope of the invention. Upon examination of the present description, additional embodiments, modifications or equivalents of the present invention will be apparent to those skilled in the art, and all such embodiments, modifications and equivalents are deemed to be included in the present invention.

The drawings are provided solely to assist in understanding the description and should not be construed in any way as limiting the scope of the invention. The drawings show the following:

Fig. 1 schematically illustrates the dependence of the PPG signal shape on the wavelength and user.

Fig. 2A schematically illustrates the difference in the mechanisms of scattering and absorption of radiation in the volume under study in the transmission mode and in the reflection mode; fig. 2B schematically illustrates the shapes of the PPG signal in transmission mode and reflection mode.

Fig. 3A schematically illustrates the absorption spectra of whole blood at different hemoglobin concentrations; fig. 3B schematically illustrates the absorption spectra of tissue (blood + other components) at the same hemoglobin concentration.

Fig. 4A schematically illustrates a first conventional approach to processing a PPG signal.

Fig. 4B schematically illustrates the second traditional approach to processing a PPG signal.

Fig. 5A illustrates the effect of using the first traditional approach to PPG signal processing in relation to determining the hemoglobin concentration of FIG. 4A.

Fig. 5B illustrates the effect of using the second traditional approach to processing the PPG signal in relation to determining the hemoglobin concentration in FIG. 4B. In fig. 5B, the term “critical point” means a set of critical points for the waveform of a signal at each of the specified wavelengths.

Fig. 6A, B illustrate a block diagram and a schematic diagram of a wearable device with a hemoglobin concentration detection function.

Fig. 7 illustrates a flowchart of a method for determining hemoglobin concentration.

Fig. 8 schematically illustrates the stage of converting a PPG signal from the time to frequency domain.

Fig. 9 illustrates the key steps of a method for determining hemoglobin concentration.

Fig. 10 illustrates a key feature of the conversion step of FIG. 7.

Fig. 11A, B illustrate the features of converting a PPG signal into the frequency domain and a PPG signal in the time domain under different absorption conditions.

Fig. 12 illustrates the features of choosing a reference PPG signal in the frequency domain.

Fig. 13 illustrates the features of signal analysis in the frequency domain.

Fig. 14 illustrates the principle of database compilation.

Fig. 15 illustrates the final step of the method for determining hemoglobin concentration.

Fig. 16 illustrates the results of predicting (determining) the hemoglobin concentration in the blood of a user of a wearable device using a database.

Detailed Description of Embodiments of the Invention

The present invention is a wearable device with the function of determining the concentration of hemoglobin in the user's blood, containing at least one PPG sensor configured to irradiate the user's tissue, including skin, bones, blood, blood vessels, with radiation of at least two different wavelengths and detect at least two PPG signals at said at least two different wavelengths in reflection mode, wherein the wearable device is configured to convert said at least two PPG signals from the time domain to the frequency domain or time-frequency domain, describe converted PPG signals relative to harmonics to obtain a set of characteristics, determine the concentration of hemoglobin in the user’s blood based on the obtained set of characteristics and database information.

According to an embodiment, the wearable device may further comprise a housing, a processor, a battery, and a storage device housed in the housing.

According to an embodiment, the wearable device is configured to perform a description of PPG signals, which includes selecting the most significant signal from all PPG signals and using it as a reference signal; extracting harmonics from the reference signal and using them as coordinates; decomposition of all PPG signals along the mentioned coordinates to obtain a set of characteristics; in this case, determining the hemoglobin concentration involves comparing the resulting set of characteristics with sets of characteristics from the database, each of which corresponds to a specific value of hemoglobin concentration.

According to an embodiment, said at least one PPG sensor of the wearable device comprises at least one radiation source comprising two or more light-emitting diodes.

According to an embodiment, the light emitting diodes may be laser LEDs.

According to an embodiment, said at least one PPG sensor of the wearable device comprises at least one radiation detector, preferably a broadband photodetector or photodiode.

According to an embodiment, one of said different wavelengths emitted by the at least one PPG sensor of the wearable device is in the range of 495-570 nm, and is preferably 530 nm.

In an embodiment, the PPG signals provided by the wearable device are back-reflected and/or back-scattered from the user's tissue.

According to an embodiment, the wearable device is further configured to filter out motion artifacts from the PPG signals. It should be noted that frequency domain signal analysis makes the hemoglobin concentration algorithm somewhat resistant to motion artifacts, provided that the frequency and duration of these artifacts do not coincide with the main signal. The frequency of artifacts may coincide with the frequency of the PPG signal, but motion artifacts are non-stationary, i.e. their characteristics change over time, therefore, during signal accumulation, it seems possible to enhance essentially stationary PPG components and suppress the contribution of motion artifacts to the overall signal spectrum.

According to an embodiment, the transformation of PPG signals from the time domain to the frequency domain comprises a Fourier transform or a Hilbert-Huang transform, and the transformation from the time domain to the time-frequency domain comprises a wavelet transform.

According to an embodiment, the wearable device is configured to select the most significant signal having the best signal-to-noise ratio, the highest amplitude, and/or the least affected by noise.

According to an embodiment, the wearable device is configured to perform harmonic extraction from a reference signal, which includes determining the coordinate of the fundamental harmonic and calculating the coordinates of other harmonics.

According to an embodiment, the processor of the wearable device is configured to perform a description of PPG signals with respect to harmonics, which includes measuring for each harmonic amplitude, curve area, peak width at half height, and/or signal-to-noise ratio to obtain a set of amplitude-phase characteristics of the harmonics.

According to an embodiment, the wearable device is configured to perform the aforementioned comparison of the resulting set of characteristics with sets of characteristics from the database using a numerical method or prediction algorithm, taking into account data from other users, sets of characteristics from the database, previously measured from other users and compared with hemoglobin concentrations , previously measured by laboratory methods. In this case, the forecasting algorithm was obtained in the process of training a neural network, and the above sets of characteristics were used in the process of machine learning of the neural network.

According to an embodiment, the storage device of the wearable device is configured to store a database.

According to an embodiment, the wearable device is further configured to correlate a determined hemoglobin concentration using a laboratory-derived hemoglobin concentration value, update the database, and refine the prediction algorithm using a machine learning algorithm or neural network.

According to an embodiment, the wearable device further comprises an input and output device configured to input user profile data into a storage device and display a specific hemoglobin concentration, and a communication module configured to exchange information with a remote server and/or cloud storage.

According to an embodiment, the wearable device is further configured to determine the concentration of oxyhemoglobin in the user's blood.

According to an embodiment, the wearable device is configured to be placed on the wrist or other parts of the body, such as a finger.

In an embodiment, the wearable device is a smart device such as a smart watch or a fitness tracker.

The present invention also relates to a method for determining the concentration of hemoglobin in the blood of a user, comprising irradiating the user's tissue with radiation of at least two different wavelengths; detection of PPG signals in the time domain at the at least two different wavelengths mentioned in the reflection mode, conversion of the detected PPG signals from the time domain into the frequency domain or time-frequency domain, description of the converted PPG signals with respect to harmonics to obtain a set of characteristics, definition hemoglobin concentration in the user’s blood, based on the obtained set of characteristics and database information.

According to an embodiment of the method, at the stage of describing PPG signals, the most significant signal is selected from all PPG signals and used as a reference signal; extract harmonics from the reference signal and use them as coordinates; all PPG signals are laid out according to the mentioned coordinates; in this case, at the stage of determining the hemoglobin concentration, the resulting set of characteristics is compared with sets of characteristics from the database, each of which corresponds to a specific value of hemoglobin concentration.

According to an embodiment of the method, irradiation is carried out using at least one radiation source, which is two or more light-emitting diodes.

According to an embodiment of the method, irradiation is carried out using at least two laser light-emitting diodes.

According to an embodiment of the method, detection of PPG signals is carried out using at least one radiation detector, preferably a broadband photodetector or photodiode.

According to an embodiment of the method, one of the at least two different wavelengths of radiation to which the user's tissue is irradiated is in the range 495-570 nm, and preferably 530 nm.

According to an embodiment of the method, the PPG signals are back-reflected and/or back-scattered from the user's tissues.

According to an embodiment, the method further includes filtering PPG signals from motion artifacts. Filtering of PPG signals from motion artifacts (body movements when walking, movements of hands, fingers, trembling, etc.) can be performed, for example, using accelerometer data from a wearable device, and is carried out for the purpose of further processing of filtered signals of higher quality.

According to an embodiment of the method, the conversion of PPG signals from the time domain to the frequency domain is carried out by the Fourier transform or the Hilbert-Huang transform, and the conversion from the time domain to the time-frequency domain is carried out by the wavelet transform.

According to an embodiment of the method, the most significant signal has the best signal-to-noise ratio, the highest amplitude and/or is least affected by noise.

According to an embodiment of the method, extracting harmonics from a reference signal includes determining the coordinate of the fundamental harmonic and calculating the coordinates of other harmonics.

According to an embodiment of the method, the description of PPG signals relative to harmonics includes measuring for each harmonic the amplitude, curve area, width at half height of the peak and/or signal-to-noise ratio to obtain a set of amplitude-phase characteristics of the harmonics.

According to an embodiment of the method, the comparison is made using a numerical method or a prediction algorithm taking into account profile data from other users, sets of characteristics from a database previously measured from other users, and hemoglobin concentrations previously measured by a laboratory method.

According to an embodiment of the method, a machine learning algorithm or learning using a neural network is used as a numerical method.

According to an embodiment of the method, the database is configured to be stored on a storage device, a remote server and/or in a cloud storage.

According to an embodiment, the method further includes correlating the determined hemoglobin concentration using a laboratory-derived hemoglobin concentration value, updating the database, and refining the prediction algorithm using a machine learning algorithm or a neural network.

According to an embodiment, the method is further adapted to determine the concentration of oxyhemoglobin in the user's blood.

The present invention further relates to a system for determining the concentration of hemoglobin in the blood of a user, comprising the above-mentioned wearable device and a remote server or cloud storage, wherein the system contains means for exchanging information between the wearable device and the remote server and/or cloud storage.

According to an embodiment of the system, the database is designed to be stored on a storage device, a remote server and/or in a cloud storage and is configured to access user data from external devices through a remote server and/or cloud storage.

According to an embodiment of the system, user profile data and the results of previous hemoglobin concentration determinations are stored on a remote server and/or cloud storage.

According to an embodiment, the system further includes means for accessing user profile data and results of previous hemoglobin concentration determinations from other electronic devices.

The proposed wearable device, method and system provide the possibility of non-invasive, continuous, reliable and motion-resistant monitoring of blood properties, in particular, determination of hemoglobin concentration and blood oxygen saturation, and are suitable for non-professional use.

Exemplary embodiments of the invention will now be described with reference to the drawings. Those skilled in the art will appreciate that the various exemplary embodiments should in no way be construed as limiting the scope of the claimed invention, and that other material and technical means equivalent or similar to those listed below may be employed by those skilled in the art to perform the various operations, functions , method steps, etc. described below. This detailed description is not intended to limit the scope of the claimed invention, which is defined only by the appended claims.

In fig. Figure 1 schematically illustrates the dependence of the PPG signal shape on the wavelength and user. It can be clearly seen that the waveform of the pulse signal depends not only on the wavelength of the primary radiation, but also on the specific user. The main factors influencing the shape of the PPG signal are the absorption and scattering of radiation in the tissue, the hydrodynamics of blood vessels, and mechanical deformation in the tissue. In this case, for example, the concentration of hemoglobin in the patient’s blood is determined by the absorption of radiation in the tissue, or more precisely, as will be discussed below, by the pulsating (AC) component of absorption in arterial blood, and the contribution made to the signal due to the scattering of radiation in the tissue is hydrodynamics blood vessels and mechanical deformation is a source of error and must be eliminated. Taking into account the peculiarities of the propagation of optical radiation in biological tissue, which will be discussed below, the above processes (sources of errors) lead to a significant change in the shape of the pulse wave depending on the wavelength, and the waveform is specific to the optical properties of the tissue of each individual user. Thus, to correctly measure the hemoglobin concentration in the blood (determined by the absorption of radiation in arterial blood), it is necessary to analyze the shape of the PPG signal. The important points here are the following: the need for spectroscopic measurements at multiple wavelengths, the need to take into account that at some wavelengths the PPG signal can be very weak, the signal can have a form difficult to process, the need to measure a weak PPG signal that has a complex shape.

In fig. Figure 2A schematically illustrates the difference in the mechanisms of scattering and absorption of radiation in the tissue volume under study in the transmission mode and in the reflection mode. In the transmission mode used in medical and laboratory applications, radiation from the light source to the light detector at two different wavelengths λ ₁ and λ ₂ travels approximately the same optical path through the patient's tissue and interacts with the same areas of tissue. In the reflection mode used in wearable devices, such as smart watches and fitness bracelets, the radiation from the light source to the light detector at two different wavelengths λ ₁ and λ ₂ travels a different path and interacts with different areas of the patient's tissue. For example, radiation with wavelength λ ₂ interacts only with the patient's skin and partially with muscle tissue, including capillary blood vessels, while radiation with wavelength λ ₂ interacts with the patient's skin, with muscle tissue, including not only capillary blood vessels, but also venous and arterial vessels. However, correct hemoglobin measurement requires an equal tissue volume of interest, which is difficult to achieve in reflectance mode. Therefore, a solution needs to be found to overcome this problem.

In fig. Figure 2B schematically illustrates the shapes of the PPG signal at wavelengths of 530 nm and 655 nm in transmission mode and reflection mode. As noted above with respect to FIG. 2A, the optical emission path at different wavelengths in the transmission mode is approximately the same, so the PPG signal in this mode has a similar shape at different wavelengths, for example, 530 nm and 655 nm, as shown by the dotted lines in FIG. 2B. At the same time, the PPG signal in reflection mode has very different shapes at different wavelengths, as shown by the solid lines in Fig. 2B. Taking into account the above, the main problems of spectral analysis of the PPG signal are the following:

- The AC component of the pulse signal has a complex shape, that is, it is not a harmonic or sinusoidal signal;

- the shape of the AC component depends on the wavelength of the radiation, that is, the shape of the pulse signal carries information, including about the spectral properties of the tissue;

- limited operation in reflection mode.

When looking for a solution, it is important to take into account the significantly different optical paths for different wavelengths, which is less important in the transmission mode and critical for the reflection mode (for example, for smart watches), the need to measure the same tissue area, the different pulse shape due to differences in scattering and absorption of radiation. Thus, for the reflection mode, it is preferable to analyze the shape of the PPG signal.

In fig. Figure 3A schematically illustrates the absorption spectra of whole blood at different hemoglobin concentrations, obtained using photoplethysmography (PPG) at certain radiation wavelengths. The dashed lines represent continuous absorption spectra of whole blood at two different hemoglobin concentrations, and the circles represent absorption measured at several discrete points. From fig. 3A shows that under ideal conditions, when there is whole blood in the cuvette, the hemoglobin concentration can be determined by measuring the transmittance of optical radiation at any one wavelength. However, the whole blood case is ideal, and in real cases it is not possible to perform measurements on whole blood in a non-invasive manner.

In fig. Figure 3B schematically illustrates the absorption spectra at the same hemoglobin concentration, but in the presence in one case of other blood components or other tissues, including melanin, lipids, proteins, etc. From fig. 3B shows that at some wavelengths, for example, at λ _m , noticeable absorption of optical radiation occurs due to the presence of other components of blood or other tissues. Thus, other blood components or other tissues may absorb at the same wavelengths as hemoglobin. Therefore, it is impossible to say definitively whether the absorption at wavelength λ _m has changed due to hemoglobin or something else.

As can be seen from Fig. 3A and B, the ideal case in blood testing is to measure blood absorbance only, where absorbance at any wavelength corresponds to changes in blood hemoglobin concentration. In actual practice, the absorption of entire tissue, including blood, is measured, and other components may also absorb radiation at the same wavelengths, affecting the pulse signal. Therefore, measuring absorption at a single wavelength may lead to incorrect results. Therefore, in this case, a spectroscopic approach to measurements is necessary, i.e. To correctly determine the hemoglobin concentration, it is necessary to measure the absorption spectrum at several wavelengths.

In fig. 4A schematically illustrates the first traditional approach to processing a PPG signal, which consists of measuring the area under the curve (AUC) of the AC component of the PPG signal, i.e. determining the integral area of the PPG signal shape. It is believed that the intensity of the AC component measured in this way is proportional to the optical transmittance of the measured volume of tissue and can be converted into absorption in this area according to the Bouguer-Lambert-Beer law. This is the simplest measurement approach, applied in transmission mode as discussed above. It is obvious that those shown in FIGS. 4A signals have approximately the same integral area, but different curve shapes. Since this approach does not take into account information about the pulse shape, the measurement of optical transmission and absorption in the reflection mode is carried out with a large error.

In fig. 4B schematically illustrates the second traditional approach to processing the PPG signal, which consists of estimating the shape of the PPG signal curve. The first peak of the pulse wave, corresponding to the anacrotic period of the pulse wave, is formed during systole (the “systolic peak” is designated SIS). The amplitude value of the anacrotic phase is also called the amplitude of the pulse wave and corresponds to the stroke volume of blood during cardiac output, thus providing indirect information about the degree of inotropic effect. The second peak of the pulse wave, corresponding to the dicrotic period of the pulse wave, is formed due to the reflection of blood volume from the aorta and large great vessels and partially corresponds to the diastolic period of the cardiac cycle (“diastolic peak” is designated as DIA). The dicrotic phase provides information about vascular tone. The top of the pulse wave corresponds to the largest volume of blood, and its opposite part corresponds to the smallest volume of blood in the tissue area under study. On the descending front of each wave there is a noticeable depression - the dicrotic incisura, which corresponds to the closure of the aortic valve (the "dicrotic incision" is designated DN). The nature of the pulse wave depends on the elasticity of the vascular wall, pulse rate, volume of the tissue area being examined, width of the lumen of the vessels, etc. It is believed that the frequency and duration of the pulse wave depends on the characteristics of the heart, and the size and shape of its peaks depend on the condition of the vascular wall. If the pulse wave clearly shows the systolic and diastolic peaks, as well as the valley between them (the upper PPG signal in Fig. 4B, a rare case of an ideal curve, characteristic of young, healthy people with a strong cardiovascular system), an approach to determining the optical properties tissue according to the shape of the PPG signal curve works well. In most cases, these features of the pulse wave are not clearly expressed (the middle graph in Fig. 4B, typical for relatively healthy adults); this approach works in many cases, but is prone to errors, since it requires individual adjustment of the algorithm for each curve, which means it is difficult to optimize processing. However, in some patients, the systolic and diastolic peaks, and the trough between them, may be difficult to detect or even indistinguishable (bottom graph in Fig. 4B, a rare case of a poor curve, characteristic of older people with weak cardiovascular systems or young unhealthy people ). In this case, the shape of the wave signal curve is characterized by the same parameters that have a clear physiological meaning, as described above, but the formalization of the algorithm for finding these points is difficult and therefore the shape is characterized inaccurately.

Additionally in FIG. 5A illustrates the effect of using the first traditional approach (AUC determination) to processing the PPG signal of FIG. 4A using the example of determining hemoglobin concentration. In fig. 5A, dashed curves show the absorption spectra of whole blood with different concentrations X and Y of hemoglobin, that is, the dependence of absorption on the radiation wavelength; circles indicate absorption values determined from the areas of the PPG signal, that is, from the integral areas of the AC components of the signal. At the same time, the circles for different concentrations on the presented graphs are located close to each other and are noticeably shifted from the true absorption values. This means that determining the hemoglobin concentration from the areas of the AC component of the PPG signal results in an incorrect concentration value. Moreover, with this approach it is often impossible to distinguish between two different blood hemoglobin concentrations.

Regarding the use of the second traditional approach to processing the PPG signal in relation to determining the hemoglobin concentration using the critical points of the pulse signal, as shown in FIG. 5B, in one case, at hemoglobin concentration Y (upper curve), the approach allows for fairly accurate measurements, however, in other cases (lower curve), it is generally impossible to determine radiation absorption at some wavelengths due to the inability to distinguish critical points on the PPG signal curve, and This means that it is impossible to determine the hemoglobin concentration from the shape of the pulse signal curve.

Unlike the approaches discussed, the method of the present invention can be applied to any pulse waveform, as will be described below.

Next, with reference to FIG. 6A, 6B will show a schematic diagram of a wearable device 201 with a function for detecting the hemoglobin concentration in the user's blood.

With reference to FIG. 6A, a block diagram of a preferred embodiment of a wearable device 201 will be described, comprising a housing 202, a processor (microprocessor) 203, a battery 204, at least one PPG sensor 208, an input device 209, an output device 210, a display 211, a storage device 212, and a communication module 213 including a wireless communication module 223 and a wired communication module 224. The input device 209 and the output device 210 may be configured as a single input and output device. The wearable device 201 may include additional sensors, such as inertial motion sensors, gyroscopes, accelerometers, an electrocardiogram (ECG) measurement sensor, an atmospheric pressure sensor, a humidity sensor, a Hall sensor, an ambient light sensor, etc. In embodiments, the wearable device 201 may be a smart device, in particular, a smart watch or fitness bracelet, a medical wearable device for monitoring health status with the function of determining hemoglobin concentration, etc. The wearable device is an electronic wearable device and can be configured with the ability to establish a wired or wireless communication channel with external devices, for example, devices 225, 226, such as smartphones, wearable fitness bracelets, voice assistants, smart TVs, smart watches, etc., or a server 227 and with the ability to transmit data via network 228 or cloud storage 229.

The wearable device 201 with the PPG sensor 208 in reflection mode is capable of detecting light backscattered and/or backreflected from the user's tissues, including skin, bones, blood, and blood vessels.

Wearable devices with a PPG sensor use reflection mode to conveniently place the device, for example, on the user's hand. When worn on the arm or other part of the user's body, the PPG sensor is in contact with the skin. However, a PPG sensor measuring in reflective mode can be affected by motion artifacts and pressure fluctuations. Any movement, such as physical activity, can lead to motion artifacts that distort the PPG signal and limit the accuracy of physiological parameter measurements. Pressure fluctuations acting on the sensor, such as the contact force between the PPG sensor and the measurement site, that is, the force with which the device is pressed against the user's skin, can deform the geometry of the artery due to compression. Thus, in the reflection mode, the amplitude of the variable component of the PPG signal can be influenced by the pressure exerted on the skin. Such influences on the measurement of the PPG signal are described in more detail in the article by Toshiyo Tamura, Current Progress of Photoplethysmography and SpO ₂ for Health Monitoring, Biomedical Engineering Letters (2019) 9, cc. 21-36 (https://doi.org/10.1007/s13534-019-00097-w).

The electrical circuit of a PPG sensor may contain an amplifier, a high-pass filter (about 0.1 Hz) to cut off the DC component and obtain pulsating changes in the signal, a low-pass filter (about 30 Hz) to eliminate high-frequency noise, and a microprocessor. The frequency range used depends on the circuit design. The PPG sensor may have a wireless module for transmitting data to an external device. In the preferred embodiment of the present invention illustrated in FIG. 6B, the wearable device 201 is in the form of a smart watch that is designed to be placed on the user's wrist. In other embodiments of the present invention, the wearable device 201 may be configured to be placed on other parts of the user's body, such as a finger.

In fig. 6B, the wearable device 201 is illustrated in a top view and a bottom view. According to this embodiment, the body 202 of the wearable device 201 includes a first surface 214 (or front surface), a second surface (or rear surface) 215, and a side surface 216 surrounding the space between the first surface 214 and the second surface 215, and fastening members 217 for for detachably attaching the wearable device 201 to the wrist, for example, using a watch strap.

According to an embodiment of the present invention, the processor 203 is configured to receive the at least two PPG signals from the at least one PPG sensor 208, store them in the storage device 212, and load them from the storage device 212 to perform processing operations.

The processor 203 is configured to filter the at least two PPG signals with different wavelengths from motion artifacts, separate the at least two PPG signals with different wavelengths into AC and DC components of the time series, and transform these time series (time domain signal) into the frequency domain or time-frequency domain and then extracting their amplitude-phase harmonic characteristics, wherein the transformation of said time series into the frequency domain may comprise a Fourier transform or a Hilbert-Huang transform, and the transformation of said time series into a frequency domain the time domain may contain a wavelet transform or a bilinear time-frequency distribution.

The processor 203 is configured to receive additional data (eg, data from said additional sensors of the wearable device 201), store it on the storage device 212 of the wearable device 201, and load it from the storage device 212 to perform operations, if necessary.

The processor 203 of the wearable device 201 is configured to transmit information about the hemoglobin concentration in the user's blood, determined based on the readings of the sensors 208, to the output device 210 to inform the user.

The processor 203 is configured to load data received by the communication module 213 into the storage device 212, and/or load data from the storage device 212 into the communication module 213 for transferring it to an external device (for example, devices 225, 226, server 227, or cloud storage 229).

The battery 204 is configured to supply power to at least one component of the wearable device 201.

With reference to FIG. 6B illustrates that the at least one PPG sensor of the wearable device 201 is located on its rear surface. Thus, during operation, the radiation surface of the PPG sensor comes into contact with the user’s wrist. The wearable device 201 may include multiple PPG sensors.

Said at least one PPG sensor 208 is configured to measure at least two PPG signals with different wavelengths. In the embodiment of the present invention shown in FIG. 6B, the at least one PPG sensor 208 is located on the rear surface 215 side of the housing 202 (the surface facing the user's skin). In one embodiment of the present invention (not shown), the wearable device 201 may include two PPG sensors 208, each of which contains one radiation source 221 and one radiation receiver 222, and the radiation sources 221 of these two PPG sensors emit light of different wavelengths. . In this embodiment of the present invention, two PPG sensors 208 are capable of capturing two PPG signals with different wavelengths. In the embodiment of the present invention shown in FIG. 6B, the wearable device 201 includes one PPG sensor 208 containing two radiation sources 221 and two radiation receivers 222, these radiation sources 221 emitting light of different wavelengths. In this embodiment of the present invention, a single PPG sensor 208 is capable of acquiring two PPG signals of different wavelengths. In an embodiment of the present invention, the radiation source 221 may comprise a light-emitting diode (LED). Photobarriers may be provided between radiation sources and/or receivers to reduce noise and eliminate crosstalk.

Input device 209 is configured to receive data that can be used by another component (eg, processor 203) of wearable device 201 inputted from outside (eg, a user of wearable device 201). In one embodiment of the present invention, input device 209 may be configured to input a user profile. The processor 203 of the wearable device 201 may receive the user profile from the input device 209, store the user profile in the storage device 212, read it from the storage device 212, and use it as needed. The wearable device 201 may also provide voice input and output.

User profile data may include the user's gender, age, height, weight. Also, the input and output device may include a screen for displaying information, for example, a certain hemoglobin concentration. The screen may be configured to use an on-screen keyboard for data entry.

The output device 210 may be configured to display information to the user on a first side 214 of the wearable device 201. The output device 210 may include a display 211 that may be configured to display output information and located on the first side 214 of the wearable device 201. Additionally, the display 211 may be configured to use an on-screen keyboard to enter data, such as a user profile.

Storage device 212 in a preferred embodiment may be configured to store:

databases with sets of amplitude-phase characteristics of harmonics compared with hemoglobin concentrations,

user profile data,

sets of measurement data containing amplitude-phase characteristics of harmonics and including:

at least two PPG signals with different wavelengths measured by said at least one PPG sensor 208,

dates of collection of each set of measurement data;

collection times of each set of measurement data,

hemoglobin concentrations determined by the processor for each set of measurement data, etc.

Also, memory 212 may store various additional data, such as data collected from additional sensors of wearable device 201. In addition, memory 212 may store various instructions that, when executed on processor 203, cause processor 203 to control components of wearable device 201. associated with the processor 203, and perform various data processing or calculations.

A communication module 213 may be used to transmit data from or to the wearable device. According to an embodiment, communications module 213 may include a wireless communications module 223 (e.g., a cellular communications module, a short-range wireless communications module, or a global navigation satellite system (GNSS) communications module) or a wired communications module 224 (e.g., a local area network communications module (LAN) or power line communication (PLC) module). Thus, the communication module 213 may support establishing a wireless communication channel between the wearable device 201 and an external device (eg, device 225, device 226, server 227, or cloud storage 229) and performing communication through the established communication channel. Communications module 213 may include one or more communications processors that operate independently of processor 203 and support direct (eg, wired) communications or wireless communications. A respective one of these communication modules may communicate with an external electronic device via a first network 228 (e.g., a short-range communication network such as Bluetooth™, wireless (Wi-Fi), or infrared data transmission (IrDA)) or a second network 228 (eg, a long-distance communications network, such as a cellular network, the Internet, or a computer network, such as a local area network or wide area network (WAN)). These various types of communication modules may be implemented as a single component (eg, a single chip) or may be multi-component (eg, multiple chips).

Next, with reference to FIG. 7, a flowchart of a method for determining hemoglobin concentration according to a preferred embodiment will be described. The claimed method can be carried out using the claimed wearable device with the function of determining the concentration of hemoglobin in the user’s blood.

First, at step 710, the user's tissue is irradiated with a light source emitting light of at least two different wavelengths. Next, at step 720, PPG signals from the user's tissue at the at least two different wavelengths are detected. Thereafter, at step 730, the detected PPG signals are converted from the time domain to the frequency domain. From all the PPG signals available at multiple wavelengths, the most significant signal with the best signal-to-noise ratio, highest amplitude, and least affected by noise is selected at step 740 and used as a reference signal. Then, at step 750, harmonics are extracted from the reference signal and used as coordinates, along which the rest of the PPG signals are laid out at step 760 to obtain a set of amplitude-phase characteristics of the signal harmonics. Finally, at step 770, the hemoglobin concentration is determined, wherein the determination of the hemoglobin concentration is carried out by comparing the resulting set of characteristics with sets of characteristics from the database, each of which corresponds to a specific hemoglobin concentration value. In this case, the database is compiled in advance based on the obtained sets of characteristics for a plurality of users of the wearable device, and these sets of characteristics are correlated with specific hemoglobin values measured by a laboratory method.

The steps of the proposed method will now be described in more detail.

Irradiation of the user's tissue at at least two wavelengths is accomplished using a light source, such as an LED(s). One of these at least two different wavelengths has a range of 495-570 nm, that is, the LED emits green light, since it is usually green light that provides the most informative signal when determining, for example, hemoglobin concentration. A reflectometric method is used, that is, the user's tissue is irradiated and then PPG signals from the user's tissue are detected at the at least two different wavelengths mentioned in the reflection mode. In the example illustrated in FIG. 8, signals are detected at six wavelengths. However, the number of PPG signals at different wavelengths is not particularly limited. The detected signals are presented in the time domain (time series). In general, several representations of a signal are possible: in the time domain, when f(t) is expressed as a function of time, in the frequency domain, when the spectrum (i.e., the amplitudes of the various harmonics) is defined, or in the time-frequency domain. Signals can be converted from time to frequency/time-frequency domain using various types of transformations. According to an embodiment of the invention, the tool for representing the detected signal in exponential terms is the Fourier transform.

However, the invention is not limited to the use of the Fourier transform and other transforms such as the Hilbert-Huang transform, wavelet transform, bilinear time-frequency distribution can be used. The function F(o) is the direct Fourier transform of the signal f(t). F(o) represents the signal f(t) in the frequency domain. The temporal representation defines a certain signal at each moment in time, while the frequency representation characterizes the relative amplitudes of the frequency components of the signal, i.e. harmonics

When performing a Fourier transform based on the characteristics of a signal in the time domain, it is possible to obtain its amplitude-phase characteristics of harmonics (instantaneous spectrum, spectral density, etc.). For example, the Fourier transform allows each PPG pulse to be converted into a set of amplitudes and phases, so the resulting set of features contains all the information necessary for the machine learning model used to determine the hemoglobin concentration. In this case, for calculations, it is convenient to describe a set of signal characteristics using complex numbers. Any of these representations completely defines the signal. However, as was explained in detail in the “Background of the Invention” section, processing the PPG signal in the frequency domain avoids a number of disadvantages of the prior art.

Conversion of signals from the time domain to the frequency domain can also be done by the Hilbert-Huang transform, and to the time-frequency domain by the wavelet transform or bilinear time-frequency distribution.

The Hilbert-Huang transform (HHT) is an empirical mode decomposition (EMD) method for nonlinear and non-stationary processes, i.e. decomposition into amplitude-time components, and subsequent Hilbert spectral analysis (HSA). NNT is a time-frequency analysis of signals and does not require an a priori functional basis for the transformation. The basis functions are obtained adaptively directly from the data by screening procedures for “empirical mode” functions. Instantaneous frequencies are calculated from the derivatives of phase functions by the Hilbert transform of basis functions. The main informative features of the Hilbert-Huang transform are the amplitude, instantaneous frequency and phase of the empirical modes.

Wavelet transform (WT) is an integral transform that is the convolution of a wavelet function with a signal. The wavelet transform converts a signal from a time representation to a time-frequency representation and is a generalization of spectral analysis, a typical representative of which is the classical Fourier transform. The term “wavelet” translated from English means “small (short) wave.” Wavelets are a generalized name for families of mathematical functions of a certain form, which are local in time and frequency, and in which all functions are obtained from one base (generating) by means of its shifts and stretches along the time axis. Wavelet transform algorithms make it possible to take into account local changes in PPG signals and carry information in three-dimensional format - amplitude, frequency, time. Typically, wavelet transforms are divided into discrete (DWT) and continuous (CWT). DWT is used for signal conversion and coding, CWT is used for signal analysis. Wavelet transforms are now increasingly being used for a huge variety of applications, often replacing the conventional Fourier transform.

Bilinear time-frequency distributions or quadratic time-frequency distributions arise in a subfield of signal analysis and processing called time-frequency signal processing, as well as in the statistical analysis of time series data. Such methods are used in cases where the frequency composition of the signal can change over time. Such analysis was formerly called time-frequency signal analysis and is now more commonly called time-frequency signal processing due to advances in the use of these techniques to solve a wide range of signal processing problems. Compared to other time-frequency analysis methods, such as the short-time Fourier transform, the bilinear transform may not have greater clarity for most practical signals, but it provides an alternative basis for exploring new definitions and new methods. All bilinear distributions are mutually convertible to each other.

At the bottom of FIG. Figure 8 illustrates PPG signals at two different wavelengths in the time and frequency domains.

The present invention allows the precise pulse waveform to be characterized from PPG signals, including weak and/or complex signals. The key steps of the method for determining hemoglobin concentration, illustrated in FIG. 9 are:

- conversion of detected PPG signals from the time domain (time series) into the frequency (or time-frequency) domain;

- selection from all detected PPG signals at different wavelengths in the frequency domain of the most significant signal, which is characterized by the best signal-to-noise ratio, the highest amplitude and which is least affected by noise, and using the selected signal as a reference signal;

- extracting the fundamental harmonic (fundamental component) from the reference signal and, on its basis, other harmonics of the reference signal and using them as coordinates along which the rest of the PPG signals are laid out, that is, the amplitude-phase characteristics of the harmonics are obtained.

In fig. 10 illustrates a key feature of the conversion step of the method of FIG. 7, consisting in the fact that when converting a PPG signal from the time to the frequency domain, information about the periodicity of the time signal is transferred to the position of the first harmonic of the frequency domain, and information about the shape of the signal (peaks, valleys, notches) is transferred to harmonics of higher orders. Ultimately, data sets are obtained for each wavelength of the pulse signal in the frequency domain, each of which is characterized by specific characteristics (amplitudes a ₁₀ , a ₁₁ ...a _1n at harmonic frequencies f ₀ , f ₁ ...f _n ), as shown in the lower part fig. 9. In this case, the position, number, amplitude, width of the resulting harmonics, which are easily measured characteristics, completely describe the pulse wave in the frequency domain. Thus, converting PPG signals into the frequency domain makes it possible to extract the amplitude-phase characteristics of the signal harmonics.

Taking into account the above, processing PPG signals in the frequency domain leads to the following advantages compared to processing in the time domain:

- a finite number of characteristics instead of continuous time series data;

- an unambiguous connection between the characteristics and the shape of the PPG signal;

- the presence of a new coordinate system (harmonic frequencies), determined from one of the signals, which allows all signals to be decomposed uniformly according to a single coordinate system.

If you try to analyze the pulse waveform by critical points, then each signal will have to be processed independently. In this case, the position of the critical points may “float”, thereby disrupting the unambiguity in the description of the signal shape at different wavelengths. In addition, in the case of weak signals and signals with a high level of noise, finding critical points may in principle be impossible. Thanks to the above-mentioned coordinate system (frequencies and their positions), it is possible to find the characteristics (for example, amplitudes) of these frequencies even in the spectrum of a signal with a high noise level.

In fig. The features of converting a PPG signal into the frequency domain are illustrated. Shown here is a high-quality reference PPG signal at wavelength λ ₂ and a regular low-quality PPG signal at other wavelengths λ ₁ , λ ₃ , etc. After converting the high quality signal into the frequency domain, the position and characteristics of the fundamental harmonic f ₀ can be easily measured. With a similar conversion of a low-quality signal, it is impossible to reliably determine the position of the fundamental harmonic f ₀ in the frequency domain. Moreover, the intense false peaks, shown in the graph on the right with broken circles, hide the necessary characteristics. However, the advantage of the present invention is the stability, or more precisely, the independence of the reference PPG signal from interference and low-quality signals. After performing the Fourier transform, the fundamental harmonic f ₀ of the signal in the frequency domain is the first harmonic corresponding to the periodicity of the PPG signal, i.e. heart rate. In the case of other time-frequency transformations, the main component is, for example, the frequency component of the greatest power and the most stable over time, which is not directly related to physiological parameters. Moreover, the position of the fundamental harmonic f ₀ of a high-quality signal in the frequency domain corresponds to the position of the fundamental harmonic f ₀ of low-quality signals.

In fig. 11B illustrates another feature of the PPG signal in the time domain, which affects the need to select an appropriate wavelength for reference measurements. According to the prior art, the best PPG signal is obtained using green light (~530 nm). However, this is not always the case. The pulse signal strongly depends on the color of a person’s skin, his nationality, tanning, skin pigmentation, etc. In FIG. Figure 11B shows PPG signals for light-skinned and dark-skinned people. The choice of a PPG signal of higher quality is based, as a rule, on determining the highest signal-to-noise ratio. As can be seen from Fig. 11B, for a fair-skinned person, the best PPG signal is actually obtained in green light, and it is used as a reference signal. However, dark-skinned people have their own characteristics, the main of which is related to the fact that green light penetrates to a shallow depth into the patient’s tissue. Because green light in dark skin is almost completely absorbed by melanin, it does not reach the blood vessels. At the same time, red/IR light penetrates deeper into the tissue, since melanin absorbs red/IR light much less than green light and does not affect the PPG signal. Therefore, for dark-skinned people, the best results will be obtained by using the PPG signal obtained using red/IR light as a reference signal, rather than green light, as for light-skinned people.

Thus, individual settings are necessary, and the appropriate reference wavelength is preferably determined for each wearable device user individually.

In particular, in FIG. 12 illustrates as an example two sets of PPG signals in the time and frequency domains at different wavelengths, and in one case, the signal obtained using green light is selected as the reference signal, characterized by the best signal-to-noise ratio, highest amplitude and low noise , for example, with wavelength λ ₃ , and in another case - red light, for example, with wavelength λ ₅ .

In fig. 13 further illustrates the features of signal analysis in the frequency domain. A signal with wavelength λ ₂ was selected as a reference PPG signal. In the top graph of Fig. Figure 13 shows the characteristics of the first f ₀ and second f ₁ harmonics of the PPG signal at wavelength λ ₂ , namely the amplitudes a ₂₀ and a ₂₁ of these harmonics. Let us note once again that in the frequency domain the positions of the harmonics are identical for signals of all wavelengths of the visible radiation spectrum. In other words, the positions of the harmonics of the analyzed PPG signal with wavelength λ ₁ , shown in the lower graph, coincide with the positions of the harmonics of the reference PPG signal with wavelength λ ₂ . Consequently, PPG signals at all wavelengths can be analyzed equally and simultaneously as soon as the harmonics are determined from the reference PPG signal. In this case, the positions of higher order harmonics are uniquely related to the positions of the fundamental harmonic by the following relationships: etc.

Thus, knowing the positions of the harmonics of the reference signal, you can easily determine the positions, and therefore the amplitudes, of the corresponding harmonics of the analyzed PPG signal with wavelength λ ₁ , for example, the amplitudes a ₁₀ and a ₁₁ of the first f ₀ and second f ₁ harmonics.

Using the method described above, based on the position of the fundamental harmonic f ₀ of the reference PPG signal, the characteristics, in particular, the amplitudes of all PPG signals at other wavelengths are determined. The advantage of this technique is the simple parameterization of the PPG signal waveform simultaneously at all radiation wavelengths.

In fig. 14 illustrates the principle of database compilation for non-invasive, personal and/or on-demand monitoring of blood properties such as hemoglobin concentration. In the present invention, this is achieved as follows. As a result of machine learning of a neural network, a prediction algorithm is created in advance. To train a prediction algorithm to determine (predict) hemoglobin concentration, a database was compiled. To do this, for each of the many users of a wearable device, using a PPG sensor installed in the wearable device, a set of amplitude-phase characteristics of harmonics of PPG signals in the frequency domain was determined etc.). Additionally, for each of the many users, the hemoglobin concentration (Hb ₁ , Hb ₂ , etc.) was determined by the reference method, for example, by performing a blood test using any of the laboratory methods. After this, the corresponding sets of PPG signal characteristics determined using a wearable device were correlated with hemoglobin concentrations determined using a laboratory blood test, obtaining a database in which a set of PPG characteristics ₁ corresponded to the hemoglobin concentration Hb ₁ , a set of PPG characteristics ₂ corresponded to the hemoglobin concentration Hb ₂ and etc. The sets of characteristics obtained using the wearable device and the hemoglobin concentrations determined by the laboratory method were recorded in a database. Thus, the database contains all possible combinations of hemoglobin concentration with PPG signal characteristics, that is, many sets of PPG signal data, each of which is associated with a specific hemoglobin concentration. The database compiled in this way was used for machine learning of the forecasting algorithm. The prediction algorithm used in the processor of the wearable device determines the hemoglobin concentration based on sets of characteristics obtained by processing PPG signals for a specific user of the wearable device.

In fig. 15 illustrates the final step of a non-invasive method for determining hemoglobin concentration. At this stage, the user uses a wearable device with a PPG sensor to monitor the concentration of hemoglobin in the blood. First, using a wearable device, a set of PPG signal characteristics is measured for the user, then they access the database and, using a prediction algorithm or a numerical method, determine in it a set of PPG signal characteristics that are most similar to the characteristics of the measured signal. Since the characteristics of the measured signal and the database-determined signal are substantially the same, the hemoglobin concentration corresponding to the database signal will also correspond to the measured signal. For example, as shown in FIG. 15, the data set of the measured PPG signal coincides with the PPG ₂ data set, and therefore corresponds to the hemoglobin concentration Hb ₂ .

Thus, to determine the hemoglobin concentration using the claimed method:

1. PPG signals are detected at several wavelengths from a user with an unknown hemoglobin concentration and converted to the frequency domain.

2. The amplitude-phase characteristics of the harmonics are extracted, describing the PPG signals at several wavelengths: PPG {a ₁₀ , a ₂₀ ... a _mn }.

3. Compare the resulting set of characteristics with the sets of characteristics of FIG ₁ , FIG ₂ ... FIG _n from a pre-compiled database, as shown above.

4. Determine the hemoglobin concentration.

In fig. 16 illustrates a graph of the results of predicting the hemoglobin concentration in the blood of a wearable device user from a database containing information on 840 tests performed on 170 test subjects. When collecting reference data, a laboratory analysis of capillary blood was used using a colorimetric method based on the hemoglobinase assay, through which the concentration of hemoglobin in the blood of the test subject was determined.

In the graph, the X-axis shows the hemoglobin concentration of the test subject, measured by a laboratory method, and the Y-axis shows the hemoglobin concentration predicted using the prediction algorithm. The X=Y line shows the correlation. In this case, a machine learning regression model using gradient descent and k-fold cross-validation were used to evaluate the prediction performance.

The accuracy characteristics of predicting hemoglobin concentration using one of the models used were: MAE 1σ=0.67, R=0.66; where MAE (Mean Absolute Error) is the mean absolute error, which is a measure of errors between paired measurements of the same phenomenon, that is, in this case, between the actual values of hemoglobin concentration measured by a laboratory method and the predicted values obtained using a wearable device; MAE is calculated as the sum of absolute errors divided by the sample size; 1a indicates that only those measurements are taken into account whose error lies within one standard deviation from the average error value over all points; R Pearson correlation coefficient.

The data obtained clearly demonstrate that the hemoglobin concentration in the blood of a test subject can be predicted based on at least a set of measurement data taken from the sensors of a wearable device, with high accuracy relative to the readings obtained by the laboratory method. As an example, one forecasting model currently has a MAE 1σ of 0.67 g/L. It is obvious that the accuracy of forecasting will increase both as the forecasting algorithm is refined and as information with the results of measurements on tested subjects is accumulated in the database. Thus, it is expected that using the wearable device according to the invention it will be possible to determine the hemoglobin concentration with high accuracy (within 1%).

The solution to the problem of predicting the concentration of hemoglobin in the user's blood was based on identifying empirical patterns in the training data placed in a common database using a machine method.

In the above-described embodiment of the present invention, the prediction algorithm is created for a wearable device located on the wrist, since the training of the prediction algorithm was based on data captured by the sensors of the wearable device located on the wrist. In other embodiments of the present invention, it is possible to train a prediction algorithm based on the placement of the wearable device on other parts of the body, such as a finger. The prediction algorithm described herein may be implemented as software including one or more instructions that may be executed by the processor 203 of the wearable device 201 or by external processors/processing devices.

The above-described training process of the prediction algorithm used profile data of other users (test subjects), measurement data sets previously measured from other users, and reference data sets previously measured by a laboratory method, as described in detail above.

As discussed below, the database can be continuously updated and, in addition, the prediction algorithm is refined through machine learning.

In the process of machine learning, methods for selecting and processing characteristics and corresponding discrete frequencies can be used:

- in the Relief-F method, the vector of weights of features (characteristics) is calculated and normalized, and then features are selected whose weight exceeds the value of a given threshold;

- the Correlation-based Feature Selection (CFS) method combines an evaluation formula with an appropriate correlation measure and a heuristic search strategy;

- the Fast Correlation Based Filter method begins to work with the full set of features, uses a measure of symmetric uncertainty to determine dependencies between features and allows you to select a subset by searching and sequentially eliminating uninformative features;

- the Sequential Forward Feature Selection (SFFS) method at each iteration adds to the set a feature that provides the best recognition efficiency for a given iteration; the Mutual Information method determines the nonlinear correlation dependence instead of calculating the Pearson correlation “feature-feature” and “feature-label”;

- basic machine learning (artificial intelligence) algorithms; As a working solution, a combination of these regression methods and any derivative regression methods that are based on the basic algorithm can be used:

- Decision Trees / Random Forests;

- support vector machines (Support Vector Machines);

- Linear Analysis;

- deep learning methods - artificial neural networks.

Best Option

To implement the method according to the invention, for example, a smart watch containing a PPG sensor is used. The LED(s) built into the smartwatch emits light at at least two wavelengths, which is partially absorbed in the tissues of the user's hand, and partially reflected and detected by a PPG sensor. The sensor detects light at the at least two wavelengths and a set of detected PPG signals in the time domain is obtained. Then, using the processor built into the smartwatch, which is also used for subsequent stages, a reference PPG signal at a certain wavelength is selected. The reference PPG signal is selected from among all available signals at different wavelengths based on the best signal-to-noise ratio, highest amplitude, and minimum noise (most often a signal at the green light wavelength). The PPG signals are then converted from the time domain to the frequency domain using the Fourier transform. In the frequency domain, the position of the first harmonic (fundamental component) f ₀ is extracted from the reference PPG signal and used to analyze signals at other wavelengths. Based on the position of the first harmonic f ₀ , the amplitude-phase characteristics of the harmonics of all signals are determined. Thus, for all combinations of wavelengths of the PPG signal and harmonics, a set of characteristics is obtained: a ₁₀ , a ₂₀ ... a _m0 , a ₁₁ , a ₂₁ ... a _m1 ... a _1n , a _2n .... a _mn , which uniquely describes the shape of the pulse wave at each wavelength A.

Additional embodiment 1

The present embodiment relates to the use of machine learning (ML) in the claimed method. As in the above-described embodiment, PPG signals at several wavelengths are first detected from a user with an unknown hemoglobin concentration in the blood, then characteristics describing PPG signals at several wavelengths are extracted, obtaining a set of characteristics: PPG {a ₁₀ , a ₂₀ ... a _mn }. In contrast to the above-described embodiment, a situation may arise where, when comparing the resulting set of characteristics with the sets included in the database, it will not be possible to unambiguously determine the hemoglobin concentration, since the resulting set will not uniquely correspond to any set from the database, being approximately between two sets that indicate the concentrations of Hb ₁ and Hb ₂ respectively. In this case, a laboratory analysis of the user’s blood is carried out, the result is entered into the database, correlating it with the obtained set of characteristics, and the prediction algorithm is additionally refined using the machine learning method. Additionally, using machine learning, a prediction algorithm can be trained to predict the association of available sets of PPG waveform characteristics with different hemoglobin concentrations.

The advantageous effects of using a machine learning method are to detect blood values for a user whose data is not included in the database and improve the accuracy of blood counts.

Additional embodiment 2

The present embodiment is aimed at more accurately determining a reference PPG signal. As described above, the reference signal is typically selected based on the best signal-to-noise ratio, highest amplitude, and most often a signal having a green light wavelength. Thus, if the wavelength at which a good signal is obtained is known, additional hardware can be used with the wearable device that increases the intensity of the PPG signal at a given wavelength, for example, the wavelength of green light, and defines it as a reference signal. After selecting the reference signal in this way, further steps of the method are carried out and the hemoglobin concentration is determined.

The advantageous effects of using this approach are that there is no need for a reference signal search algorithm, and that the method works for any user (regardless of his physiological characteristics, skin color, etc.).

Examples of use of the invention

The invention can be used to unobtrusively monitor the hemoglobin concentration of the user of a wearable device, such as a smart watch or fitness bracelet. By continuously monitoring a user's hemoglobin concentration, both short-term and long-term deviations from the normal hemoglobin concentration range can be detected. Short-term deviations are not critical and can be eliminated personally by the user based on available recommendations. For example, with a short-term decrease in hemoglobin concentration, which most often indicates anemia, it may be recommended to change the diet, increase physical activity, spend more time in the fresh air, etc. At the same time, long-term deviations may indicate serious health problems and the need to see a doctor. In this case, the obtained data can be sent to the doctor remotely to collect a primary medical history.

The advantage of this use of the invention is the low power consumption of wearable devices, which means a longer period of continuous monitoring, and better data analysis. It also ensures a sensitive user experience.

Another example of the use of the invention can be used by athletes. As noted above, the main function of hemoglobin is to bind and transport oxygen to internal organs. Deviation of the Hb concentration value from the norm leads to a decrease in endurance during sports. A low concentration of hemoglobin leads to oxygen starvation of the muscles (hypoxia or low number of red blood cells), resulting in: lethargy, periodic dizziness, fatigue, muscle pain, shortness of breath, rapid heartbeat. A high concentration of hemoglobin leads to dehydration or a large number of red blood cells, resulting in: lethargy, nosebleeds, fatigue, muscle pain. Such conditions are difficult to tolerate, especially during physical activity. Muscles need oxygen to get enough energy, and a lack of oxygen results in the inability of fibers to contract adequately. Athletes may not immediately notice the symptoms of anemia that has begun, which is associated with increased endurance of the body. Therefore, it is necessary to know the hemoglobin concentration before training, and it is better to constantly monitor it.

Moreover, oxygen deprivation is fraught with consequences for ordinary people, since a lack of oxygen in the blood can negatively affect brain function.

Another example of the use of the invention is assistance in choosing a training regimen, which is important for beginner athletes, athletes after a long break from training, triathletes, sportswomen and when training at altitude. An athlete using a smartwatch switches it to hemoglobin concentration measurement mode. Given continuous sensor data, the user's profile and geolocation, a machine learning algorithm analyzes the data from the hemoglobin sensor, which is then provided to the user as a hemoglobin concentration reading. Based on the controlled hemoglobin concentration, personalized training recommendations can be provided, such as mode, duration, load level, etc.

The present invention expands the functionality of the Samsung Health application in smart watches by providing another parameter for a comprehensive analysis of the user's health. Smart watches can provide the following indicators: daily fluctuations in hemoglobin concentration, current hemoglobin concentration within the normal hemoglobin concentration, continuous measurement results throughout the day, average values for a given period.

Although the invention has been described with certain illustrative embodiments, it should be understood that the invention is not limited to these specific embodiments. On the contrary, the spirit of the invention is intended to include all alternatives, modifications and equivalents that may be included within the spirit and scope of the claims.

In addition, the invention includes all equivalents of the claimed invention, even if the claims change during examination.

Claims (48)

1. A wearable device with the function of determining the concentration of hemoglobin in the user’s blood, containing: at least one photoplethysmographic (PPG) sensor configured to irradiate the user's biological tissue with radiation of at least two different wavelengths and detect at least two PPG signals at said at least two different wavelengths in reflection mode, wherein the wearable device is configured to: convert the at least two PPG signals from the time domain into the frequency domain or time-frequency domain, select from all PPG signals the most significant signal, characterized by the highest signal-to-noise ratio, the highest amplitude and/or the least noise, and use it as a reference signal, extract harmonics from the reference signal and use them as coordinates, carry out decomposition of all PPG signals along the mentioned coordinates to obtain a set of amplitude-phase characteristics of harmonics and determine the concentration of hemoglobin in the user’s blood by comparing the resulting set of amplitude-phase characteristics of harmonics with sets of amplitude-phase characteristics of harmonics from the database, each of which corresponds to a specific value of hemoglobin concentration. 2. A wearable device according to claim 1, wherein said at least one PPG sensor comprises at least one radiation source comprising two or more light-emitting diodes. 3. A wearable device according to claim 1 or 2, in which said at least one PPG sensor contains at least one radiation detector, preferably a broadband photodetector. 4. A wearable device according to any one of paragraphs. 1-3, wherein one of the various wavelengths mentioned is in the range of 495-570 nm, and preferably 530 nm. 5. A wearable device according to any one of paragraphs. 1-4, wherein the wearable device is additionally configured to filter out motion artifacts from PPG signals. 6. A wearable device according to any one of paragraphs. 1-5, wherein the transformation of PPG signals from the time domain to the frequency domain contains a Fourier transform or a Hilbert-Huang transform, and the transformation from the time domain to the time-frequency domain contains a wavelet transform. 7. A wearable device according to any one of paragraphs. 1-6, wherein extracting harmonics from the reference signal involves determining the coordinates of the fundamental harmonic and calculating the coordinates of other harmonics. 8. A wearable device according to any one of paragraphs. 1-7, while the decomposition of PPG signals into coordinates includes measuring for each harmonic the amplitude, curve area, width at half height of the peak and/or signal-to-noise ratio to obtain a set of amplitude-phase characteristics of the harmonics. 9. A wearable device according to any one of paragraphs. 1-8, while the comparison is based on the use of a numerical method or prediction algorithm taking into account profile data of other users, sets of amplitude-phase characteristics of harmonics from the database, previously measured from other users and compared with hemoglobin concentrations previously measured by a laboratory method. 10. A wearable device according to any one of paragraphs. 1-9, further comprising a storage device configured to store a database. 11. The wearable device according to claim 9 or 10, wherein the wearable device is further configured to correlate a determined hemoglobin concentration using a laboratory-derived hemoglobin concentration value, update the database, and refine the prediction algorithm using a machine learning algorithm or neural network. 12. A wearable device according to any one of paragraphs. 1-11, additionally containing: an input and output device configured to input user profile data and display a determined hemoglobin concentration, and a communication module configured to exchange information with a remote server and/or cloud storage. 13. A wearable device according to any one of paragraphs. 1-12, wherein the wearable device is further configured to determine the concentration of oxyhemoglobin in the user's blood. 14. A wearable device according to any one of paragraphs. 1-13, wherein the wearable device is configured to be placed on the wrist. 15. A wearable device according to any one of paragraphs. 1-14, wherein the wearable device is a smart device, preferably a smart watch or fitness bracelet. 16. A method for determining the concentration of hemoglobin in the user’s blood, including the steps of: irradiate the user's biological tissue with radiation of at least two different wavelengths, detecting photoplethysmographic (PPG) signals in the time domain at said at least two different wavelengths in reflection mode, convert the detected PPG signals from the time domain into the frequency domain or time-frequency domain, select the most significant signal from all PPG signals, characterized by the highest signal-to-noise ratio, the highest amplitude and/or the least noise, and use it as a reference signal, harmonics are extracted from the reference signal and used as coordinates, all PPG signals are laid out along the mentioned coordinates to obtain a set of amplitude-phase characteristics of harmonics and determine the hemoglobin concentration in the user's blood by comparing the resulting set of amplitude-phase characteristics of harmonics with sets of amplitude-phase characteristics of harmonics from the database, each of which corresponds to a specific value of hemoglobin concentration. 17. The method according to claim 16, in which the irradiation is carried out using at least one radiation source, which is two or more light-emitting diodes. 18. The method according to claim 16 or 17, in which the detection of PPG signals is carried out using at least one radiation detector, preferably a broadband photodetector. 19. Method according to any one of paragraphs. 16-18, wherein one of the at least two different wavelengths is in the range 495-570 nm, and preferably 530 nm. 20. Method according to any one of paragraphs. 16-19, additionally including filtering PPG signals from motion artifacts. 21. Method according to any one of paragraphs. 16-20, in which the conversion of PPG signals from the time domain to the frequency domain is carried out by the Fourier transform or the Hilbert-Huang transform, and the conversion from the time domain to the time-frequency domain is carried out by the wavelet transform. 22. Method according to any one of paragraphs. 16-21, in which extracting harmonics from a reference signal involves determining the coordinate of the fundamental harmonic and calculating the coordinates of other harmonics. 23. Method according to any one of paragraphs. 16-22, in which the decomposition of PPG signals into coordinates includes measurement for each harmonic of the amplitude, curve area, width at half height of the peak and/or signal-to-noise ratio to obtain a set of amplitude-phase characteristics of the harmonics. 24. Method according to any one of paragraphs. 16-23, in which the comparison is carried out using a numerical method or prediction algorithm taking into account profile data of other users, sets of amplitude-phase characteristics of harmonics from the database, previously measured from other users and compared with hemoglobin concentrations previously measured by a laboratory method. 25. The method according to claim 24, in which a machine learning algorithm or learning using a neural network is used as a numerical method. 26. Method according to any one of paragraphs. 16-25, wherein the database is configured to be stored on a storage device, a remote server and/or in a cloud storage. 27. Method according to any one of paragraphs. 24-26, further including correlating a determined hemoglobin concentration using a laboratory-derived hemoglobin concentration value, updating the database, and refining the prediction algorithm using a machine learning algorithm or a neural network. 28. Method according to any one of paragraphs. 16-27, additionally designed to determine the concentration of oxyhemoglobin in the user’s blood. 29. A system for determining the concentration of hemoglobin in the user’s blood, containing a wearable device according to any one of paragraphs. 1-15 and a remote server and/or cloud storage, wherein the system contains means for exchanging information between the wearable device and the remote server and/or cloud storage. 30. The system according to claim 29, wherein the database is designed for storage on a storage device, a remote server and/or in cloud storage and is configured to access user data from external devices through a remote server and/or cloud storage. 31. The system according to claim 29 or 30, wherein user profile data and the results of previous determinations of hemoglobin concentration are stored on the storage device, remote server and/or cloud storage. 32. The system of claim 31, further comprising means for accessing user profile data and the results of previous hemoglobin concentration determinations from other electronic devices.