Method for determining blood pressure data

Abstract

The invention relates to a measuring device and methods for determining blood pressure data using a PPG sensor and an opto-mechanical tissue model. Abstract figure: Figure 5

Description

Title of the invention: Method for determining blood pressure data

[0001] The present description relates to a PPG device, a PPG system and a method for determining blood pressure data, including arterial blood pressure data, using an optical sensor, such as a photoplethymosgram (PPG) sensor. Context

[0002] The use of PPG sensors to determine blood pressure has been a subject of research for many years. However, the relationship between blood pressure and PPG is complex and multifactorial. Numerous attempts and research projects have been conducted, but the lack of a reliable blood pressure measurement device using a PPG sensor on the market illustrates the need for new methods. Summary

[0003] The method and device described in this application use optical data obtained from a user (in a tissue) and a model of light propagation within that tissue. Since this optical data is representative of a local deformation of the tissue due to the blood pulse, the optical data is also referred to as PPG data.

[0004] According to one embodiment, the description relates to a method for determining blood pressure data, the method comprising: - obtaining measured photoplethysmogram (PPG) data from a user, - determining blood pressure data using the measured PPG data and a hybrid opto-mechanical model.

[0005] In particular, the PPG data is obtained from a PPG signal measured by a PPG sensor. The PPG sensor can be mounted on a measuring device that also includes control circuitry. The method can be implemented by the control circuitry.

[0006] In one embodiment, the blood pressure data are differential pressure data. This method is particularly effective when combined with a hybrid linear opto-mechanical model.

[0007] The hybrid opto-mechanical model is a combination of a mechanical model of the tissue and an optical model of the light inside that tissue.

[0008] In one embodiment, the hybrid opto-mechanical model (the mechanical model of the latter) includes a poroelastic model. More precisely, this The poroelastic model can be linear. In this case, as previously stated, the blood pressure data can be differential pressure data.

[0009] In one embodiment, the hybrid opto-mechanical model (the optical model of the latter) includes a scattering model. The scattering approach implies a high scattering / absorption ratio.

[0010] In one embodiment, the determination of blood pressure data includes, using a lookup table between simulated blood pressure data and simulated PPG data calculated using the hybrid opto-mechanical model: - the extraction of at least one characteristic of the measured PPG signal, - the identification of a simulated PPG signal closest to the measured PPG signal using at least one extracted feature, and - the return of simulated blood pressure data associated with the nearest simulated PPG as determined blood pressure data.

[0011] According to another aspect, the disclosure relates to a method for determining differential pressure data, the method comprising: - obtaining measured photoplethysmogram (PPG) data from a user, - determining differential pressure data using the measured PPG data and a linear mechanical model.

[0012] In one embodiment, the model is a hybrid opto-mechanical model.

[0013] The additional features disclosed for the previous aspect also apply to this aspect.

[0014] The disclosure also relates to a computer program product comprising instructions which, when executed by a control circuitry of a measuring device, cause the measuring device to execute a method as described above.

[0015] The disclosure also relates to a measuring device comprising an optical sensor (for example a PPG sensor) and control circuitry, wherein the control circuitry is configured to implement a method as described above, and wherein the PPG data are processed from a PPG signal measured by the optical sensor.

[0016] The measuring device can be a watch, an armband, a finger device, etc. Presentation of the drawings

[0017] The following drawings are provided for illustrative purposes only: - [Fig. 1]: [Fig. 1] illustrates a schematic representation of a measuring device with an optical sensor, - [Fig.2]: [Fig.2] illustrates several types of measuring devices as schematically represented in [Fig.1], - [Fig.3]: [Fig.3] illustrates a schematic representation of a measurement device and its connected ecosystem, - [Fig.4]: [Fig.4] shows equations of the model, - [Fig. 5]: [Fig. 5] shows the breakdown of the different stages of the model. - [Fig. 6]: [Fig. 6] shows the PPG signal, the derivative and the second derivative, notably used to determine the characteristics of the PPG signal, called PPG data, - [Fig. 7]: [Fig. 7] shows a comparison of the simulated data (obtained with the model) and the measured data, - [Fig.8]: [Fig.8] shows a correspondence between the measured differential pressure values ​​and the simulated differential pressure value identified for the nearest neighbor, - [Fig.9]: [Fig.9] shows a diagram illustrating a method according to a mode of disclosure implementation. Detailed description

[0018] Figure 1 illustrates a measuring device 100 comprising an optical sensor 102 configured to measure a user's optical data. The optical sensor includes one or more light emitters 104 (for example, one or more LEDs), configured to emit light towards the user, and a light receiver 106, configured to receive the light emitted by the light emitter 104 and reflected (or scattered) by the user's tissues. More precisely, only a localized area of ​​the user receives light. This localized area is typically the tip of a finger, part of the wrist, an earlobe, etc. The measuring device 100 generally includes a housing 108, on which or through which the optical sensor 102 is mounted. A finger of the user 110 is shown interacting with the optical sensor 102.

[0019] In one embodiment, the optical sensor 102 is a PPG sensor. Examples of PPG sensors are provided in [Fig.2], with a watch 202 (such as the Withings™ ScanWatch™, an Apple Watch™, a Garmin™ watch, etc.), a finger sensor 202 (known as a finger pulse oximeter), a multiscope device 204, such as the Withings™ BeamO™, an armband, etc.

[0020] The localized area is perfused by blood vessels, in which the blood circulates in a pulsatile manner. The blood contains hemoglobin, which absorbs light, so that a variation in the hemoglobin content induces a variation in the light received.

[0021] The PPG sensor can operate at several wavelengths. For example, the wavelength can be in the green, red, and infrared ranges, for example below 600 nm or 540 nm, between 500 nm and 600 nm, and between 800 and 1000 nm. PPG sensors are known per se and will not be disclosed in further detail.

[0022] The PPG 102 sensor generates a measured PPG signal, which can then be processed to generate measured PPG data.

[0023] In one embodiment, the measuring device 100 is a connected device forming part of an ecosystem 300 shown in [Fig. 3], which can exchange data with a computing device. The computing device can be a smartphone or a server.

[0024] For example, the measuring device can communicate with a mobile terminal 314 (smartphone type) and / or a remote server 316. In one embodiment, the measuring device 100 communicates with the mobile terminal 314 (via BLE for example), which itself communicates with the server 316 (via WIFI for example).

[0025] The measuring device 100 includes a control circuit 302 with a processor 304, a memory 306, and an I / O ("input / output") interface 308 configured to send and receive data to and from the control circuit 302. A communication module 310 is designed to exchange data with an external device (e.g., a smartphone). The communication module 310 can be a wireless module, such as Wi-Fi, Bluetooth, Bluetooth Low Emission, etc. The control circuit 302 is connected to the optical sensor 102 to exchange data, including sending instructions to it and receiving optical data from it. The control circuit 302 is further configured to process the optical data. To display the data to a user, the measuring device 100 may include a display 311.

[0026] The measuring device may include a battery for storing energy.

[0027] Memory 306 is configured to store instructions which, when executed by processor 304, cause the measuring device to execute one of the methods described herein. The methods can be executed locally by process 304 of the measuring device. No connection to the external device is required to execute any of these methods.

[0028] The measuring device 100 can communicate, using the communication module 310 and a communication network 312, with an external device, such as a mobile terminal 314 or a server 316.

[0029] The communication network can be of different types: short-range wireless (Bluetooth, Wi-Fi, etc.), long-range wireless (cellular data, 3G, 4G, 5G, etc.), wired (Ethernet, etc.).

[0030] Hybrid opto-mechanical method, device and model

[0031] The measuring device 100 is configured to determine blood pressure data using the measured PPG data obtained by means of the optical sensor 102 and using a hybrid opto-mechanical model. The PPG data originates from a PPG signal obtained by the PPG sensor 102.

[0032] In one embodiment, the blood pressure data includes the differential pressure, which is the difference between systolic and diastolic pressure. Using the differential pressure instead of blood pressure simplifies tissue modeling and the propagation of scattered light within it.

[0033] Model description

[0034] The hybrid opto-mechanical model comprises two models: one for the propagation of light, called the optical model, and the other for the tissues, called the mechanical model, which takes into account the deformation of the tissues and their mechanical properties.

[0035] In one embodiment, the mechanical model is a poroelastic model. More precisely, in one embodiment, the mechanical model is a linear poroelastic model. To take full advantage of this model, the blood pressure data here includes differential pressure data. In another embodiment, the mechanical model is a nonlinear poroelastic model. This model allows the calculation of systolic and / or diastolic pressures.

[0036] Poroelastic model

[0037] The assumption of this poroelasic model is that the tissue undergoes small deformations relative to a reference configuration.

[0038] The poroelastic modeling is illustrated in [Fig. 4], with a set of 400 equations. θof is the fluid phase (β), assumed to be Newtonian (for a linear model only), and vf is the fluid (blood) velocity, where θos = l - θof is the solid phase, assumed to be elastic and linear, and νus is the displacement of the solid mixture. The mixture of the solid and fluid phases is assumed to be incompressible, but the model can be generalized. P is the Lagrange multiplier, and k is the hydraulic conductivity. The equations define a set of different partial equations in the domain (here, a volume represents the finger tissue).

[0039] Hook 402 represents the solid part of the model, hook 404 the fluid part, while hook 406 represents the conservation of momentum and support 408 the conservation of mass.

[0040] Time discretization was performed using a backward Euler method for the fluid and structure, and space discretization was performed using a Pl-Pl-(Brezzi-Pitkaranta)-stabilized method.

[0041] The document “Barré, M., Grandmont, C. and Moireau, P., 2022. Analysis of a linearized poromechanics model for incompressible and nearly incompressible materials. Evolution Equations and Control Theory” provides further information on this type of modelling.

[0042] One embodiment of the optical model for the absorption and scattering of light is a scattering model, which implies a high scattering / absorption ratio. A quasi-static scattering approximation approach has been adopted. However, a radiative transfer equation could be used, but it requires more computational resources.

[0043] The model can be broken down into several steps, illustrated in diagram 500 of [Fig. 5] with the corresponding equations. At 502, us, vf, and p are calculated (taking into account the boundary conditions); at 504, the porosity is calculated, taking into account the linearization around μ); at 506, for a given wavelength λ, the absorption coefficient is determined; at 508, the fluence μ) is found using the diffusion equation, which is then used to determine a fluence; at 510, the signal trace is calculated, using the fluence. This signal trace corresponds to a PPG signal.

[0044] These two models are combined, in particular using a finite element technique, but other techniques are possible.

[0045] To calculate the model, the software used is FELiScE ("Finite Elements for Life Sciences and Engineering"), developed by the COMMEDIA laboratories of INRIA.

[0046] Model verification

[0047] The accuracy of the model was verified using a dataset of 21 measurements (fingertip PPG signals and brachial blood pressure, on 21 individuals) and a simulation for N=72000 PPG signals, all within a predefined range for all poroelastic and optical parameters. In particular, 10 parameters were modified for the N simulations, 5 of which relate to the poroelastic part of the model and 5 to the optical part of the model (porosity, permeability, hematocrit viscosity, differential pressure, Young's modulus of the tissue, absorption of oxygenated hemoglobin, absorption of deoxygenated hemoglobin, absorption of bloodless tissue, reduced diffusion of the skin, saturation (arteriovenous).

[0048] The boundary conditions are also taken into account. In addition, a conversion of the differential pressure at the fingertip and the brachial blood pressure is carried out using the Branco and aZ mode. (Watanabe, SM, Blanco, PJ and Feijôo, RA, 2013. Mathematical model of bloodflow in an anatomically detailed arterial network of the arm. ESAIM: Mathematical Modelling and Numerical Analysis, 47(4), pp.961-985).

[0049] To compare the simulated PPG signals and the measured PPG signals, it is necessary to extract characteristics. These characteristics can include amplitudes, such as the OS amplitude, as illustrated in [Fig. 6], with a PPG signal, its derivative and its second derivative, and points marked by letters.

[0050] The results are presented in [Fig. 7], for an OS amplitude of the PPG signal (comparison between simulated and measured PPG signals) and three wavelengths of the optical sensors 102 (green, red, infrared). Points 702 are calculated for the N signals (curve 704 is the density of the calculated data), the larger points represent the OS amplitude for the measured PPG signal and the mean of the measured PPG signal (curve 706 is the density of the measured data and curve 708 the mean of the measured data). The superposition of the densities shows the relevance of the model.

[0051] Model inversion

[0052] However, once the hybrid opto-mechanical model has been constructed, an inversion can be calculated to determine a specific output based on an input. The hybrid opto-mechanical model is capable of generating a PPG signal when blood pressure data is input (among other inputs, such as mechanical and optical parameters). However, the measuring device 100 is capable of generating PPG signals. Therefore, to obtain a measuring device capable of determining blood pressure data, the hybrid opto-mechanical model must be inverted so that blood pressure data can be determined from an input PPG signal.

[0053] First, the model must show that it can be reversed, i.e. that the data relating to blood pressure can be deduced from a measured PPG signal.

[0054] Figure 8 illustrates two graphs, 802 and 804, showing the feasibility of inverting the hybrid opto-mechanical model. All the simulated data can be found in the vicinity of a characteristic (PPG data) extracted from a measured PPG signal (averaged over all heartbeats) with a side length of a certain tolerance. When several simulations exist, the closest one is chosen. Graph 802 shows the results for the green, red, and infrared wavelengths, for which, with a relative error of 20%, 5 out of 20 intersections are empty. The plot has a mean of 0.2 mmHg and a standard deviation of 0.9 mmHg.

[0055] Figure 804 uses the same reasoning for red and infrared wavelengths (green is not considered). With a tolerance of 5%, only 1 intersection out of 20 is empty. The tracing pressure has an average of 0.15 mmHg and a standard deviation of 0.8 mmHg. This shows that the model can be inverted.

[0056] There are various techniques for inverting a model or obtaining equivalent results. One of these involves creating a lookup table. Several blood pressure data points are fed into the hybrid opto-mechanical model, which generates corresponding simulated PPG data. A complete dataset is then simulated to create a lookup table between the simulated PPG data and the simulated blood pressure data. More specifically, the lookups can relate to features of the extracted PPG signals, such as those disclosed previously.

[0057] Since the model inputs include blood pressure (or the corresponding differential pressure) and other physical parameters, calibration of these physical parameters may be necessary. One way to do this is to use blood pressure to help calibrate the physical parameters and to vary only the blood pressure.

[0058] Method for determining data relating to PB

[0059] A method for determining BP-related data will now be disclosed.

[0060] The hybrid opto-mechanical model has been disclosed previously. This hybrid opto-mechanical model produces blood pressure data when PPG data is input. More specifically, the blood pressure data is the differential pressure.

[0061] Figure 9 illustrates a method 900. Method 900 can be executed by the control circuitry 302 of the measuring device 100, so that the measuring device is able to directly return the blood pressure data. In this respect, the control circuitry 302 can store the lookup table described above.

[0062] In step 902, the control circuitry 302 receives a PPG signal from a user interacting with the measuring device 100. In step 904, the measured PPG signal can be processed to generate PPG data, referred to as measured PPG data. For example, processing in step 904 may include feature extraction as described previously.

[0063] At step 904, the control circuitry 302 thus obtained measured PPG data.

[0064] In particular, the use of the hybrid opto-mechanical model means the use of the lookup table calculated using the hybrid opto-mechanical model. The control circuitry retrieves the associated simulated blood pressure data corresponding to the simulated PPG data that correspond (“ matching") to the measured PPG data. This simulated blood pressure data becomes the blood pressure data for the measured PPG data.

[0065] In step 906, the control circuitry determines the blood pressure data using the measured PPG data and the hybrid opto-mechanical model. In particular, the use of the hybrid opto-mechanical model means using the lookup table calculated using the hybrid opto-mechanical model. The control circuitry 302 identifies at least one feature in the measured PPG data and finds the closest simulated PPG data using at least one extracted feature. The closest match can be determined using an appropriate measurement (distance, etc.). The control circuitry 302 then retrieves the simulated blood pressure data associated with the simulated PPG data closest to the measured PPG data. This retrieved simulated blood pressure data becomes the blood pressure data for the measured PPG data.

[0066] In step 908, the determined blood pressure data is stored by the control circuitry 302 or displayed on the measuring device 100. The determined blood pressure data can also be sent by the control circuitry 302 to the remote computer device.

Claims

Demands

1. Method (900) for determining blood pressure data, the method comprising: - obtaining (902, 904) measured photoplethysmogram,PPG data from a user, - determining (906) blood pressure data using the measured PPG data and a hybrid opto-mechanical model.

2. Method according to claim 1, wherein the data relating to blood pressure are data relating to differential pressure.

3. Method according to any one of claims 1 to 2, wherein the hybrid opto-mechanical model is a combination of a mechanical model of the tissue and an optical model of the light inside that tissue.

4. Method according to any one of claims 1 to 3, wherein the hybrid opto-mechanical model comprises a poroelastic model.

5. Method according to claims 2 and 4, wherein the poroelastic model is linear.

6. Method according to any one of claims 1 to 5, wherein the hybrid opto-mechanical model includes a diffusion model.

7. A method according to any one of claims 1 to 6, wherein the determination (906) comprises, using a lookup table between simulated blood pressure data and simulated PPG data calculated using the hybrid opto-mechanical model: - the extraction of at least one feature in the measured PPG signal, - the identification of a simulated PPG signal closest to the measured PPG signal using at least one extracted feature, and - the return of the simulated blood pressure data associated with the closest simulated PPG as determined blood pressure data.

8. Product computer program comprising instructions which, when executed by a control circuitry (302) of a measuring device (100), cause the measuring device to execute a method according to any one of claims 1 to 7.

9. 11 Measuring device (100) comprising a photoplethysmogram sensor, PPG, (102) and a control circuitry (302), wherein the control circuitry is configured to perform a method according to any one of claims 1 to 7 and wherein the PPG data are processed from the PPG signal measured by the optical sensor (102).

10. The measuring device of claim 9, wherein the measuring device is a watch.

Images