Non-invasive sensors and measurement methods

Abstract

A method for measuring a parameter of interest in a target environment (2) by a non-invasive sensor (1) based on photoacoustic or photothermal detection, comprising: a) providing a sensor having an adaptation module (14) having a processor executing an inverse modeling algorithm; b) the adaptation module selecting an initial illumination model setting; c) the adaptation module determining an optimal illumination case in a correspondence table; d) a light source irradiating the target environment according to the optimal illumination case; e) the detection cell detecting a signal; f) the processor of the adaptation module returning a current model setting (CMmes) and an estimate of the parameter of interest (Pest); g) the processor of the adaptation module evaluating the selected illumination model setting, and only if this illumination model setting is different from the current model setting, g1) the adaptation module receiving the current model setting (CMmes) and returning a new model setting, after which c), d), e), and f) are repeated; and h) the measured value of the parameter of interest (Pmes) by the sensor is the final estimate of the parameter of interest (Pest).

Description

[Technical field]

[0001] The present invention relates to process measurement and non-invasive sensors that enable the measurement of one or more parameters of interest in a target environment.

[0002] More precisely, the invention relates to a non-invasive sensor based on the detection of the photothermal or photoacoustic effect, in particular adapted to measure a parameter in a target environment, such as a layered and / or changing environment, the parameter being measured being for example blood glucose in the skin. [Background technology]

[0003] In the field of biosensors, it is known to realize non-invasive sensors based on photoacoustic or photothermal detection.

[0004] A zone of interest in the environment to be analyzed, called the target, is illuminated by a laser beam of wavelength and modulation frequency chosen according to the parameter of interest to be measured. The laser beam is absorbed in the target to a depth determined by the structuring of the target. The energy-absorbing light then locally heats the target. In response to this heating, a thermal wave is generated in the target, with a frequency equal to the laser modulation frequency. This wave propagates through the target, in particular to the outer surface of the target.

[0005] Thermal waves can be detected and analyzed directly. Now, photoacoustic detection takes advantage of the fact that thermal waves are associated with the same pressure wave frequency as the modulation frequency.

[0006] Indirect photoacoustic detection involves detecting pressure waves generated in the fluid external environment when a thermal wave generated in the target propagates to the target-fluid external environment interface.

[0007] This photoacoustic effect has been the subject of many theoretical studies. Allan Rosencwaig and Allen Gersho, among others, have developed a theoretical model of the photoacoustic signal, which includes the physicochemical properties of the sample being analyzed, such as the optical diffusion length, the thickness, and the thermal diffusion length (Rosencwaig, A. and Gersho, A. (1976), Theory of the photoacoustic effect with solids, Journal of Applied Physics, 47, 64).

[0008] Hu et al. developed a theory generalized photoacoustic effect in layered materials (Hu, H., Wang, X., & Xu, X. (1999). Generalized theory of the photoacoustic effect in a multilayer material. Journal of Applied Physics, 86, 3953-3958).

[0009] Photoacoustic detection offers many advantages compared to other detection techniques, among which can be mentioned the orthogonality aspect of the transformation: an optical signal at the input of the analyzed environment is transformed into an acoustic signal that is highly specialized to the phenomenon being observed and allows the use of inexpensive miniature sensors.

[0010] Difficulties with photoacoustic or photothermal detection include, inter alia: A number of parameters roughly affect the detection signal, For a particular analyte of interest that is present in low concentrations in the environment being analyzed, only a portion of the detection signal is specific to each of these parameters of interest.

[0011] In layered materials, to be able to deduce from a photoacoustic or photothermal signal the concentration of an analyte of interest in a given layer, it is necessary to know all the other parameters that affect this signal: in particular the structure of the material, i.e. the thickness of the various layers that make up the material, their respective physicochemical compositions (apart from the parameter of interest to be measured) and possibly their respective thermal conductivities or the thermal resistance associated with each interface between two successive layers.

[0012] Sensors based on photoacoustic or photothermal detection can be calibrated once only (or less frequently relative to the duration of use) if only the parameter of interest changes in the target. On the other hand, if the properties of the layered environment to be analyzed change, the sensor must be calibrated periodically to obtain measurement results with an acceptable accuracy. This problem arises especially for sensors intended for use in vivo. For example, in the case of noninvasive sensors for interstitial blood glucose monitoring, noninvasive sensors based on photoacoustic or photothermal detection can only be calibrated with limited accuracy, since the composition of the skin not only differs between patients but also changes over time for a given patient.

[0013] Document EP2460470 describes a process for periodic calibration of a non-invasive blood glucose sensor that includes a near-infrared laser source. It should be noted that this sensor uses a detection method other than photoacoustic or photothermal detection, since it is part of the incident light wave that is transmitted or scattered by the substance to be detected. The technical constraints and limitations of such a sensor are therefore not the same as sensors based on photoacoustic or photothermal detection, especially with regard to energy consumption and measurement accuracy.

[0014] In the calibration process of EP 2 460 470, a number of calibration models, possibly obtained by numerical simulation, or a number of data sets generating these models are stored in the sensor memory.

[0015] During calibration, the optical spectrum of the tissue to be analyzed, called the reference spectrum, is measured by illuminating the tissue over a range of wavelengths, and based on this a calibration model for the measurements to follow is selected.

[0016] At a later date of the analysis, the optical spectrum of the analyzed tissue is measured again and the difference between the measured spectrum and the reference spectrum is calculated. Based on this difference, the quality of the preselected calibration model is evaluated and, if this evaluation is unfavorable, a new calibration model is selected for determining the measured blood glucose level. The blood glucose level is then determined by the sensor by substituting the measured absorbance at each wavelength of the spectrum in the calibration model.

[0017] By adapting the calibration model bit by bit over time according to the process of EP2460470, blood glucose can be measured with greater accuracy compared to blood glucose sensors where the calibration is performed only once when the sensor is adapted to the patient. A drawback of the process of EP2460470 is that for each recalibration, the optical spectrum of the tissue to be analysed must be acquired over the full wavelength range, which is associated with a large energy expenditure.

[0018] As a consequence, if the method of EP2460470 were to be purely and simply adapted with a continuous sensor based on indirect photoacoustic detection, the energy costs of the process would be unsustainable and would require batteries of a weight that would be unacceptable to patients wearing such sensors.

[0019] Moreover, photoacoustic detection requires not only the selection of the wavelength but also the selection of the laser modulation frequency. In fact, the penetration depth of the laser into the target is determined by the modulation frequency of the laser. If the structure of the target (e.g. skin) changes over time, the modulation frequency used to measure the parameter of interest (e.g. interstitial blood glucose) also changes over time. To carry out the calibration process of EP2460470 for a sensor based on photoacoustic or photothermal detection, it is necessary to change not only the wavelength of the laser as in the conventional case, but also the modulation frequency of the laser so that the difference between the reference spectrum and the calibrated spectrum can be calculated. In other words, it is necessary to acquire a series of optical spectra of the target at different modulation frequencies prior to each measurement (or each series of measurements). Therefore, the energy consumption problem becomes even more serious if the method of EP2460470 is simply transferred to a non-invasive sensor based on indirect photoacoustic detection.

[0020] Finally, EP2460470 preliminarily presents a skin model consisting of three layers: a superficial layer with a thickness of 0.1 mm, an internal layer with a thickness of 0.9 mm, and a subcutaneous layer with a thickness of 2.0 mm. Although EP2460470 mentions the possibility of using the (total) thickness of the skin as a variable in the simulation (paragraph 79), it does not prove the feasibility of such an embodiment, nor does it indicate whether a number of layers different from three can be considered. It therefore seems that the calibration process of EP2460470 is not able to take into account all the variability inherently present in the layered environment investigated. The present invention therefore aims to improve the accuracy of non-invasive sensors based on photoacoustic or photothermal detection while controlling their energy consumption, or to reduce the energy consumption while controlling the accuracy of these sensors, in particular in layered and / or extensive target environments. Summary of the Invention

[0021] The present invention therefore provides a method for measuring a parameter of interest in a target environment by means of a non-invasive sensor based on photoacoustic or photothermal detection, comprising the steps of: a) providing a sensor, A light source; a device for controlling the illumination parameters of the light source; a detection cell configured to detect an acoustic or thermal signal; a memory storing a correspondence table including model settings each representing a given state of a target environment and optimal irradiation cases each including a set of irradiation parameters, each model setting being associated with an optimal irradiation case; an adaptation module for exchanging information with a device for controlling the irradiation parameters of the detection cell and the light source, the adaptation module comprising a processor configured to receive as input an irradiation case including a set of irradiation parameters and an acoustic or thermal signal and to execute an inverse modeling algorithm providing as output a model setting and values ​​of the parameters of interest; providing a sensor comprising: b) an adaptation module selecting an initial irradiation model setting; c) the adaptation module determines in the correspondence table the optimal irradiation case for the selected irradiation model setting, i.e. the irradiation case that allows the measurement of the parameter of interest with a given accuracy and / or a measurement data amount that allows a minimum energy consumption; d) a light source illuminating a target environment according to a set of illumination parameters of said optimal illumination case; and e) detecting an acoustic or thermal signal generated in response to the irradiation by the detection cell; f) a processor of the adaptation module executes an inverse modeling algorithm, receiving as input the acoustic or thermal signals detected by the detection cells and the optimal irradiation case used for the irradiation, and returning as output the current model settings and estimates of the parameters of interest; g) the processor of the adaptation module evaluates the selected exposure model setting by comparison with the current model setting, and only if this evaluation is unfavorable g1) c), d), e) and f) are repeated after which the adaptation module (14) receives as input the current model setting (CMmes) and returns as output a new exposure model setting (CMirrad); h) The measurement of the parameter of interest by the sensor (Pmes) is the final estimate of the parameter of interest (Pest). The present invention relates to a measurement method including the steps of:

[0022] According to these provisions: A blood glucose measurement result can be obtained by a single irradiation (without executing sub-step g1 if the evaluation is favorable). This irradiation is performed according to an irradiation case that is selected by default while corresponding to the control of energy consumption (for example, the number of irradiation parameters selected is limited). In step g) of evaluating the irradiation model setting, it is possible to be confident that the current case was actually the highest possible accuracy in light of the irradiation cases listed in the correspondence table, which is more advantageous than a process in which step g) is not provided and this confidence cannot be obtained. Alternatively, the blood glucose measurement result can be obtained by two irradiations (step g with execution of sub-step g1 if the evaluation is unfavorable, and thus repeating c), d), e) and f) again), and the second irradiation is performed according to the irradiation case that allows the highest possible accuracy in the particular case in light of the additional information obtained by the first irradiation and the irradiation cases listed in the correspondence table. This allows the accuracy of the process to be known and improved compared to a process without step g).

[0023] Thus, the method allows in the first case to verify the selected irradiation parameters and to be confident that the measurement accuracy is the best possible for this sensor setting and the target state on the measurement day, taking into account the additional information obtained during the irradiation, and in the second case to adapt the irradiation parameters in such a way that the best possible measurement accuracy is obtained in a second step, taking into account the additional information obtained during the first irradiation, both performed with controlled energy consumption and in particular without the need for complete or frequent acquisition of the spectral irradiation or the modulation frequency. It should be noted that a correspondence table is essential for this process, which allows the association with each current model setting and allows the selection of the optimal irradiation case for the irradiation and thus the limitation of the energy consumption per irradiation.

[0024] According to various embodiments, any of the following characteristics, alone and / or in any other combination, can be envisaged:

[0025] According to one embodiment, g1) comprises repeating g) again after f), where irradiation is repeated only if the irradiation parameters used are not optimal irradiation parameters for the target state on the measurement day, and optionally only if the number of repetitions is below a predefined threshold.

[0026] According to this embodiment, it is possible to gradually converge to the irradiation parameters that allow the most accurate acquisition, especially when the repetition of the irradiations is performed at a higher frequency (characteristic frequency of the changes in the structure of the target). Here again, the energy consumption per irradiation is controlled by the fact that the number of irradiation parameters is limited per irradiation case, but even if convergence is only obtained after 3, 4, 5 or even 10 consecutive irradiations, the overall energy consumption can be controlled and in particular even below the energy consumption required for the acquisition of the complete spectrum of irradiation and / or modulation frequencies, while controlling or at the same time improving the measurement accuracy.

[0027] According to one embodiment of this process, a correspondence table is pre-generated using a processor and a database model setting including bytes (model setting, exposure case, parameters of interest) and the acoustic or thermal signals detected by the detection cells associated with each byte, and the correspondence table is stored in the memory of the non-invasive sensor.

[0028] With this configuration, the correspondence table can be obtained automatically and updated, for example, when the database model settings are expanded or when adaptation to a given patient or patient type is required.

[0029] According to one embodiment, the measurement method includes a processor learning at least one inverse modeling algorithm from a database model configuration and storing the at least one inverse modeling algorithm in a memory of the non-invasive sensor.

[0030] This arrangement allows one or more inverse modeling algorithms to be automatically trained, e.g., adapted to a model setup and a given exposure case or set of model setups and a given set of exposure cases, respectively. According to one embodiment of the method, at least a portion of the acoustic or thermal signals detected by the detection cells associated with the bytes stored in the database model setup are simulated, i.e., generated, by a computerized simulation device.

[0031] This configuration allows, for example, the generation of a large number of bites corresponding to the possible different target environment conditions and also the possible different irradiation cases much faster and at a lower cost than would be possible by experiment in real situations. If the target environment is a living tissue such as skin, bites can be generated corresponding to rare physiological situations, extreme situations, or simply situations that differ from those that a set of test patients may experience. A correspondence table generated from such a database can thus cover a much wider range of situations with a much finer granularity than would be possible without simulation, ultimately allowing the generation of optimal irradiation cases that are more adaptive in terms of consumption and / or accuracy for each model setting.

[0032] The present invention further provides a non-invasive sensor based on photoacoustic or photothermal detection configured to measure a parameter of interest in a target environment, comprising: A light source; a device for controlling the illumination parameters of the light source; a detection cell configured to detect an acoustic or thermal signal; a memory storing a correspondence table including model settings each representing a given state of the target environment and optimal irradiation cases each including a set of irradiation parameters, each model setting being associated with an optimal irradiation case; Equipped with The non-invasive sensor further comprises an adaptation module configured to exchange information with a device for controlling the illumination parameters of the detection cell and the light source, the adaptation module comprising a processor suitable for executing an inverse modeling algorithm receiving as input an illumination case comprising a set of illumination parameters and an acoustic or thermal signal and providing as output a model setting and values ​​of the parameters of interest; The adaptation module is i. selecting an initial exposure model setting; ii. determining an optimal irradiation case corresponding to the irradiation model setting in the correspondence table; iii. sending the optimal illumination case to a device for controlling illumination parameters of the light source; iv. receiving a signal detected by the detection cell; and v. determining a current model setting and an estimate of the parameters of interest based on the received detected photoacoustic or photothermal signal and the optimal irradiation case; vi. Evaluating the exposure model setup by comparison with the current model setup; vii. determining a new irradiation model setting when receiving the current model setting only if the irradiation model setting of the selected target layered environment differs from the model setting of the current target layered environment to be compared, determining a new optimal irradiation case corresponding to the new irradiation model setting from the correspondence table, and sending the new optimal irradiation case to the irradiation parameter control device, so that the light source irradiates the target layered environment according to the set of irradiation parameters of the new optimal irradiation case, and the detection cell detects a new thermal or acoustic signal generated in response to this irradiation and sends it to the irradiation parameter control device. The target layered environment according to the set of irradiation parameters of the new optimal irradiation case, and the detection cell detects a new thermal or acoustic signal generated in response to this irradiation and sends it to the processor of the adaptation module (14) configured to repeat iv, v, vi, and vii again; viii. determining a measure of the parameter of interest based on the final estimate of the parameter of interest; and The non-invasive sensor is further configured to:

[0033] This configuration allows the sensor to measure a parameter of interest in a target environment, in particular in tissue of a living body, with controlled accuracy and / or consumption.Finally, the invention relates to a computer program comprising instructions for causing the non-invasive sensor according to the preceding embodiments to carry out the steps of the process according to any one of the above-mentioned embodiments.

[0034] In the following, embodiments of the invention will be described with reference to the drawings, which will be briefly described. [Brief description of the drawings]

[0035] [Figure 1] FIG. 1 depicts an exemplary modeling of a target layered environment such as skin. [Diagram 2] FIG. 2 is a diagram illustrating the steps of the process of optimized measurement in a particular embodiment of the invention when it is desired to measure the interstitial blood glucose of a patient. [Diagram 3] 1 shows the main elements of a sensor 1 according to the invention, in this case placed in contact with a target environment 2. [Figure 4] A diagram representing the steps performed by the simulation module 15 to simulate an optoacoustic signal detected by the optoacoustic cell in response to irradiation of a modeled target layered environment using parameters of the model setting CMk according to the irradiation parameters of the irradiation case Ij. [Diagram 5] FIG. 13 illustrates how the model works in reverse. [Figure 6a] 1 shows the results of an influence analysis of variables of SHAP ("SHapley Additive exPlanations") type performed on three inverse models trained on a first "thick" group corresponding to a model setting of two-layered skin in which the upper layer modeling the stratum corneum is thicker than 18 μm, each of which was trained with signals detected after irradiation at six different modulation frequencies, the horizontal axis represents the SHAP value, each point on the diagram corresponds to a variable and to the SHAP value of an instance, the position on the vertical axis is determined by the variable, the position on the horizontal axis is determined by the SHAP value and the intensity (gray scale) color represents the value of the variable. [Figure 6b]1 shows the results of a SHAP ("SHapley Additive exPlanations") type variable influence analysis performed on three inverse models trained on a second "thin layer" group corresponding to a model setting of a two-layer skin in which the upper layer modeling the stratum corneum is less than 18 μm thick, each of which was trained with signals detected after irradiation at six different modulation frequencies, the horizontal axis represents the SHAP value, each point on the diagram corresponds to a variable and to the SHAP value of an instance, the position on the vertical axis is determined by the variable, the position on the horizontal axis is determined by the SHAP value and the intensity (gray scale) color represents the value of the variable. [Figure 6c] 1 shows the results of a SHAP ("SHapley Additive exPlanations") type variable influence analysis performed on three inverse models trained on the entire first and second populations, each of which was trained with signals detected after irradiation with six different modulation frequencies, with the horizontal axis representing the SHAP value, each point on the diagram corresponding to a variable and the SHAP value of an instance, the position on the vertical axis being determined by the variable and the position on the horizontal axis by the SHAP value, and the intensity (gray shade) color representing the value of the variable. [Figure 7a] FIG. 13 shows blood glucose levels predicted by a first inverse model trained on a first "thick layer" group with signals detected after irradiation with six different modulation frequencies. [Figure 7b] FIG. 6b shows blood glucose levels predicted by a second inverse model trained on the first “thick layer” group by signals detected after irradiation according to the two variables with the greatest influence on the first “thick layer” group identified in FIG. 6a. [Figure 8a] FIG. 13 shows blood glucose levels predicted by a third inverse model trained on the second "thin layer" group with signals detected after irradiation with six different modulation frequencies. [Figure 8b]FIG. 6B shows blood glucose levels predicted by a fourth inverse model trained on the second “thin layer” group using signals detected after irradiation according to the two variables with the greatest influence on the second “thin layer” group identified in FIG. [Figure 9a] FIG. 13 shows blood glucose levels predicted by a fifth inverse model trained across the two groups "thin layer" and "thick layer" with signals detected after irradiation with six different modulation frequencies. [Figure 9b] FIG. 6c shows blood glucose levels predicted by a sixth inverse model trained on the two groups "thin layer" and "thick layer" as a whole by signals detected after irradiation according to the two variables with the greatest influence on the two groups "thin layer" and "thick layer", as identified in FIG. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0036] In the drawings, identical or similar objects are designated with the same reference symbols.

[0037] The present invention relates to a non-invasive sensor 1 for one or more parameters of a target environment 2, in particular a target layered environment 2 whose structuring may change over time. The target layered environment 2 can be human or animal tissue, such as for example the skin.

[0038] In the following, the parameters to be measured are referred to as "parameters of interest".

[0039] The parameter of interest may be a physiological parameter, particularly if the layered environment is human or animal tissue.

[0040] For example, the physiological parameter to be measured is blood glucose, in particular interstitial blood glucose. It is also possible to measure the water content of a particular layer of the skin, or the lactate concentration of a particular layer. These examples are not limiting.

[0041] The sensor 1 may be wearable and allow for continuous monitoring of a parameter of interest.

[0042] The non-invasive sensor 1 may be based on photoacoustic or photothermal detection.

[0043] The measurement method is particularly suitable for improving the accuracy of non-invasive sensors based on indirect photoacoustic detection, which detect acoustic waves generated in a fluid environment (in particular gas) surrounding the target environment 2 in response to irradiation, while controlling the energy consumption or while limiting the energy consumption while controlling the accuracy. However, it is entirely possible to carry out the process for non-invasive sensors based on photothermal energy or direct photoacoustic detection, thereby obtaining one of the above two technical effects. For ease of understanding, an exemplary indirect photoacoustic detection is described in more detail below, but generalization to sensors based on indirect photoacoustic or photothermal energy is also readily made.

[0044] The non-invasive sensor 1 is represented diagrammatically in FIG. an illumination device 11 comprising a light source 11a, an intensity modulation device 11b for said light source and at least one modulation frequency control device 11c for modulating the intensity of the light emitted by the light source by the intensity modulation device 11b; at least one photoacoustic detection cell 12 for detecting a signal generated in response to illumination of the target environment 2 by the emitted light (e.g., directly or indirectly detecting thermal waves propagating through the target environment 2); a signal processing module 13 configured to receive and process data from the at least one detection cell 12; an adaptation module 14 of irradiation parameters and calibration models; a simulation module 15 mounted together with the other elements of the non-invasive sensor 1, remote or otherwise; Equipped with:

[0045] In one particular embodiment, the light source 11a emits towards the target environment 2 a laser beam whose intensity is modulated at at least one particular wavelength.

[0046] The light source 11a can in particular be a light emitting diode (LED) or a laser chip. Additionally or alternatively, the light source 11a can include a quantum cascade laser (QCL) emitting in the mid-infrared range (MIR-QCL), an ICL laser ("interband cavity laser"), an internal or external cavity laser, a GaSb laser. These examples are not limiting. The light source 11a can be selected depending on the target environment 2 and / or the parameters of interest.

[0047] The non-invasive sensor 1 may comprise multiple light sources 11a.

[0048] The non-invasive sensor 1 also detects the light source 11a and the illumination parameters of the light source (11a), in particular: a frequency at which the at least one intensity modulation device 11b modulates the intensity of the at least one light source 11a, such that the intensity of the light emitted by the light source 11a is modulated at an adjustable modulation frequency; Optionally, the wavenumber (or equivalently, the wavelength) of the light emitted by the light source 11a, and / or Optionally, the emission power of the light source 11a at a given wavelength The at least one control device 11c also includes circuitry associated with the at least one control device 11c configured to control the at least one control device 11c.

[0049] In the following, the frequency at which the intensity of a given light source 11a is modulated at a given wavelength is referred to as fmod(λ).

[0050] Multiple wavelengths can be associated with the same modulation frequency, and multiple modulation frequencies can be associated with the same wavelength. Light emitted by light source 11a at a given wavelength λ can be characterized by this wavelength, the corresponding modulation frequency, and optionally the corresponding optical power and / or other parameters such as the integration time or duty cycle that characterize the laser pulse.

[0051] The light source 11a can be intensity modulated by any known electrical or mechanical means.

[0052] Light may be emitted by light source 11a continuously or in a pulsed manner.

[0053] The light emitted by the light source 11a and modulated in intensity entering the target environment 2 propagates to and passes through the target environment 2 (phenomenon represented by solid arrows in FIG. 3). It is then gradually absorbed by the various components of the target 2 at a depth zmax determined by the structure of the target environment 2 and its physiochemical composition. This absorption of light energy causes local heating of the target environment 2. As a result, heat waves of a frequency equal to the modulation frequency of the light source propagate through the target environment 2, especially towards its surface (phenomenon represented by dotted arrows in FIG. 3). This heat wave gives rise to pressure waves of the same frequency in the gaseous environment outside the target, which propagate through this gaseous environment surrounding the target environment 2, especially through the photoacoustic detection cell 12 (phenomenon represented by dashed arrows in FIG. 3).

[0054] The detection cell 12, in the case of photoacoustic detection, comprises a chamber filled with gas (e.g. air) through which the acoustic waves propagate, and one or more sensors suitably positioned in this space, for example facing the target environment 2. This may be one or more electroacoustic sensors, for example microphones or piezoelectric transducers, configured to convert the pressure of the acoustic waves into an electrical signal.

[0055] Each electroacoustic sensor is operatively connected to a signal processing module 13 .

[0056] The signal processing module 13 may include an analog-to-digital converter configured to convert an analog electrical signal from the electroacoustic sensor into a digital signal.

[0057] The signal processing module 13 may comprise a synchronous detection device suitable for demodulating and extracting the signal of interest from the detected signal.

[0058] The signal processing module 13 optionally comprises an operational amplifier operatively connected to the analog-to-digital converter and configured to amplify the electronic signal derived from the acoustic response of the target 2 transmitted by the electroacoustic sensor.

[0059] In an exemplary embodiment, the analog-to-digital converter is operatively connected to a digital signal processor for digital signal processing.

[0060] The non-invasive sensor 1 according to the invention also comprises an adaptation module 14 of the illumination parameters and the calibration model as well as a simulation module 15. These two elements are explained in the following sections.

[0061] Adaptation Module 14 The adaptation module 14 is Exchange of information with Simulation Module 15 (described below), Receiving information as required from the detection cell 12 and / or the signal processing module 13; transmitting information to a control device 11c about at least one illumination parameter of the light source 11a (e.g., the modulation frequency at which the intensity modulation device 11b modulates the intensity of the light emitted by the light source 11a at a given wavelength, the wave number (or equivalently, the wavelength) of the light emitted by the light source 11a, the optical power of the light source 11a at a given wavelength and optionally at a given modulation frequency, etc.); A computerized device comprising at least one processor capable of performing the steps of:

[0062] The steps carried out by a non-invasive sensor 1 comprising an adaptation module 14 for measuring a parameter of interest are: Target environment 2 is the skin (and thus in this case a layered environment), the parameter of interest is interstitial blood glucose; A particular case is depicted diagrammatically in FIG.

[0063] The non-invasive sensor 1 further comprises a storage memory for the configuration database model, the correspondence table, and / or one or more inverse models, as described below.

[0064] This storage memory may be distributed among the adaptation module (14) and / or the simulation module (15) and / or may be shared with the adaptation module (14) and / or the simulation module (15).

[0065] The adaptation module 14 executes the steps of a process that allows to select in the configuration database model the model settings of the target environment 2 that are most adapted to the target environment 2 on the measurement date, to select the optimal irradiation parameters for the measurement based on the correspondence table, and to determine the most suitable (i.e. most accurate) inverse model for the calculation of the parameters measured from the signals detected on this special basis. These steps will be explained after the explanation of the steps that allow the generation of the database model settings and the correspondence table.

[0066] The model setting database and the correspondence table are generated from a simulation module 15 embedded in the sensor 1 or a remote simulation module 15. If the simulation module 15 is remote, the non-invasive sensor 1 comprises communication means for the simulation module 15 and the adaptation module 14 to exchange data.

[0067] Simulation Module 15 The simulation module 15 is a computerized device configured to generate a set of model settings each corresponding to (or modeling or describing) a particular state of the target environment 2, a set of irradiation cases of the target environment 2, and a theoretically detected photoacoustic (or, if necessary, photothermal) signal in response to each irradiation case for each model setting CMk of the target environment 2 from an analytical model of the target environment 2 and the photoacoustic (or, if necessary, photothermal) detection cell 12, as shown in FIG.

[0068] This simulation module 15 is particularly relevant when the target environment 2 changes over time and / or is layered, where the target environment 2 adopts different real settings over time that can each be modeled by a particular model setting (one or more concentrations change within one or more layers of the target environment 2, and therefore one or more dimensions change, such as for example the thickness of one of the layers of the target environment 2 and / or the number of layers of the target environment 2).

[0069] a) Multiphysics analysis model of target environment 2 The target environment 2 to be analyzed is modeled as shown in Figure 1. The target environment 2 is composed of a series of N layers that separate an external environment A and an internal environment B, and whose interfaces are assumed to be locally planar. Each layer i

number

number

number

number

[0070] The list of Level 1 parameters can be expanded if more accurate modeling is desired: in particular, the list of Level 1 parameters describing the layers of the skin can include concentrations of other components of the skin, such as fat, lactic acid, oxygen, etc.

[0071] Furthermore, for the skin example, it is conceivable that skin color, patient age, or any other anthropometric parameter may be taken into account to expand or limit the space of possible models.

[0072] The implicit parameters of the model target environment 2 (referred to as Level 2 parameters because they are not provided as inputs to the simulation module 15) can be calculated using the analytical model. For example, from the Level 1 parameters and known equations, the thermal conductivity, heat capacity, density, or absorption coefficient at each wavelength for each layer of the target environment 2 can be estimated.

[0073] The parameter of interest, which plays a specific role, is not included in the list of level 1 settings. This parameter of interest may be known or unknown depending on the stage of the process. Its value is known for carrying out a simulation using the simulation module 15, whereas it is of course unknown in the case of an actual measurement by the non-invasive sensor 1.

[0074] The number of layers N of the target environment 2 can also be a variable of the model. Moreover, in the example of the skin, N can be 1 or more depending on the physiological situation. Thus, for a particular physiological situation, the skin can be properly represented by two layers, a first layer corresponding to the stratum corneum, which can have, for example, a low glucose concentration, and a second layer corresponding to the rest of the skin, the glucose concentration of which can be matched to the interstitial glucose concentration to be measured.

[0075] In other situations, models with three or even four layers may be more suitable, in which case the water concentration in the layers may increase, for example, with the depth at which the layers are located.

[0076] Therefore, the number N of layers of the target environment 2 does not have to be constant.

[0077] In the case of skin, the external environment A is typically the atmosphere surrounding the patient, which also fills the photoacoustic detection cell.

[0078] The simulation module 15 executes a multi-physics analytical model of the layered environment 2 based on physical and / or chemical equations such as, by way of non-limiting example, the Beer-Lambert equation for light absorption and the thermodynamic equations for heat (Fourier's law and conservation laws).

[0079] A model configuration CMk of a target environment 2, where k is a positive integer, corresponds to (or is modeled as) a particular state of a given target environment 2. We assume that this particular state is correctly represented by data for N layers and the values ​​of the level-1 parameters for each layer.

[0080] For each model setting CMk, the multiphysics analytical model allows the simulation of a thermal wave generation at the layer 1 / external environment A interface (interface 1 / A) in response to irradiation by a light source 11a whose irradiation parameters, i.e., for example the wavelength λ, the modulation frequency fmod(λ) at this wavelength and the surface density of power at this wavelength, are known when the parameters of interest are better known.

[0081] As a variant, the multiphysics analytical model allows the simulation of pressure waves generated in the external environment A.

[0082] In either case, the signal obtained at the output of the processor executing the multiphysics analytical model is referred to as the "simulated response wave."

[0083] The simulated response waves may be provided as an input to a processor that runs a detection cell model.

[0084] b) Detection cell model The non-invasive sensor 1 based on photoacoustic or photothermal detection comprises a photoacoustic (or photothermal) detection cell 12 configured to detect and analyze pressure waves (or heat waves) generated in an external environment A when heat waves generated in a target environment 2 in response to irradiation reach an interface 1 / A.

[0085] The entire detection cell 12 can be analytically modeled. The multiphysics analytical model simulates a response wave that would theoretically be received at the input of the detection cell 12, which allows the detection cell model 12 to predict the output signal of the detection cell 12.

[0086] Various models are possible.

[0087] Thus, in the case of indirect photoacoustic detection, the parameters of the detection cell model 12, hereafter referred to as cell parameters, may include these dimensions (e.g., vent size, cell height, etc.), thermodynamic state parameters (temperature, air pressure, relative or absolute humidity, etc.). The photoacoustic detection cell can in particular be modeled using an RLC equivalent circuit. For example, the model described in Dehe, Alfons et al. "Tea Infineon Silicon MEMS Microphone." (2013) may be suitable.

[0088] This photoacoustic detection cell model may include a model signal processing step, executed as required by the signal processing module 13, to generate, from each simulated response wave generated by a processor executing a multiphysics analytical model, a signal that would theoretically be obtained at the corresponding output of the photoacoustic detection cell (if applicable after processing of the signal by the signal processing module 13).

[0089] The processor of the simulation module 15 may be configured to execute a photoacoustic detection cell model.

[0090] Thus, the multiphysics analytical model of the target environment 2 and the detection cell model 12 result in a global analytical model that allows prediction of the output of the detection cell 12, or, if necessary, the expected signal in the signal processing module 13, from the model settings C m of the target environment 2 and the data of the irradiation cases I j , provided that one also provides parameters of interest (which may be advantageously chosen, since it is a simulation). This is illustrated in FIG.

[0091] Thus, to summarize, the simulation module 15 receives as input the parameters of the model setting CMk of the target environment 2, i.e. the number N of layers of the target environment 2 and the level 1 parameters per layer, as well as the parameters of interest and the illumination parameters of the illumination cases Ij, the illumination cases comprising one or more wavelengths of light emitted by one or more lasers, one or more frequencies of the respective modulations of the intensity of said lasers and, optionally, the respective powers emitted by said lasers. At output, the simulation module 15 gives a theoretically expected signal at the output of the detection cell 12 or, if necessary, a theoretically processed simulated signal (called simulated output signal) expected at the output of the signal processing module 13 for the model setting CMk of the selected target environment 2.

[0092] The simulated output signal may be stored in memory in the form of a Fourier spectrum.

[0093] The bytes (model settings CMk of the target environment 2, irradiation cases Ij, parameters of interest, amplitudes and phases of the components of the simulated output signal) may be stored in a database of model settings.

[0094] Model settings database From the above, a number of model configurations CMk can be generated, possibly automatically and / or randomly, each corresponding to a set of layer numbers N and level 1 parameters, and optionally values ​​or ranges of values ​​of the parameters of interest that represent particular situations in the target environment 2 of interest.

[0095] Also, for each model setting CMk, it is possible to generate, possibly automatically and / or randomly, a number of illumination cases Ij, each corresponding to a set of illumination parameters representative of the parameters of the light source 11a used for the illumination.

[0096] Thus, an illumination case Ij may include one or more intensity modulation frequencies of one or more lasers, the wavelengths of each of these lasers, and optionally, the illumination power by each laser.

[0097] The simulation module 15 is used to calculate the amplitude and phase of each component of the simulated output signal obtained at the processor output of the simulation module 15, which executes, if necessary, the multiphysics analytical model and the sensing cell model for each model setting CMk and irradiation case Ij, and a global analytical model including the values ​​of the parameters of interest that are additionally given.

[0098] The illumination case Ij may be the same for multiple different model setups CMk and possibly multiple values ​​of the parameter of interest, or may be different for each model setup CMk and / or value of the parameter of interest. After the simulation is performed, all these model setups CMk, illumination cases, values ​​of the parameter of interest, and the associated photoacoustic signal (or photothermal signal, if desired) simulated as bytes (model setup CMk, illumination case Ij, parameter of interest, amplitude and phase of the components of the simulated output signal) can be stored in a database of model setups.

[0099] The generation of the model settings CMk and / or the irradiation cases Ij does not have to be completely random.

[0100] The generation of the model configuration CMk may be based, inter alia, on physiological considerations that restrict the space of possibilities to physiologically realistic model configurations: for example, it is possible to restrict the possible thicknesses of the first layer of the skin to the range actually observed in experiments (8 mm, 40 mm) and also to restrict the water concentration of this layer to a limited range for each thickness (the water concentration of the stratum corneum is correlated with its thickness).

[0101] The generation of the illumination cases Ij can take into account in particular the constraints of the light sources 11a available to the non-invasive sensor 1 given in wavelength and / or power, or the range of modulation frequencies related to the type of target environment 2 to be analyzed or the wavelengths related to the parameters of interest.

[0102] As a variant, the configuration database model can contain only bytes obtained experimentally in real situations (model configuration CMk, irradiation case Ij, parameters of interest, amplitudes and phases of the components of the signal actually measured) or it can simultaneously contain bytes obtained in real situations and bytes obtained by simulation.

[0103] Correspondence table From the database of model settings, an artificial intelligence model can be trained in the simulation module 15 .

[0104] After learning, the artificial intelligence model can solve the inverse problem, i.e., by knowing the simulated photoacoustic signal and the irradiation parameters of the irradiation case Ij, the parameters of interest and the model setting CMk of the target environment 2, and thus the number of layers N and the level 1 parameters, can be found, as shown in Fig. 5.

[0105] The learned model (hereafter referred to as the inverse model) can be transmitted to the adaptation module 14 and stored in the memory of this module.

[0106] Multiple different inverse models can be trained using learning games and / or different learning rules.

[0107] To the extent that multiple irradiation cases Ij may be associated with the same model setting CMk, it is also possible to use statistical analysis techniques and / or artificial intelligence to identify the irradiation case Ij that enables measurement of the parameter of interest with a desired accuracy and / or data measurement amount that may minimize energy consumption.

[0108] In particular, an analysis of the influence of the variables on the trained models ("feature importance") can be performed. At the end of this analysis, an optimized (limited number of irradiation wavelengths and modulation frequencies fmod per irradiation wavelength) irradiation case Iopt.k can be associated with each model configuration CMk.

[0109] It is possible to work per model setup or to categorize the model setups and associate an optimal exposure case Iopt.cat.k with each category of model setup.

[0110] The purpose of the correspondence table can be understood in the light of Figures 6a to 9b. The context of these figures is to measure interstitial blood glucose in the skin. In this context, the inverse model is trained on a database of model settings including bytes (model settings (CMk), irradiation cases (Ij), parameters of interest) and the acoustic or thermal signals detected by the detection cells associated with each byte.

[0111] The data from the database is a first "thick" group, corresponding to a two-layer skin model configuration in which the thickness of the upper layer modeling the stratum corneum is greater than 18 mm; A second "thin" group corresponds to a two-layer skin model setup in which the thickness of the upper layer, which models the stratum corneum, is less than 18 mm; The two groups are:

[0112] An inverse model is trained for each group with the detected signals for six different modulation frequencies. Then, an analysis of the influence of variables is performed for each of the two inverse models obtained, as shown in Figures 6a and 6b. From such an analysis of the influence of variables, it can be concluded that: For the model setting of the “thick layer” group, the irradiation parameters (wavelength 1034 cm -1 , two different modulation frequencies (50 Hz and 200 Hz) are sufficient to obtain the desired accurate blood glucose measurement results, and analysis of the module (equivalent to amplitude) of the signal detected at each of these frequencies is sufficient. For the model setting of the "thin layer" group, the irradiation parameters (wavelength 1034 cm -1 , two different modulation frequencies 50 Hz and 400 Hz) and an analysis of the module (corresponding to the amplitude) of the signal detected at each of these frequencies is sufficient to obtain a blood glucose measurement result with the desired accuracy.

[0113] It is therefore understood that in this process, the adaptation of the exposure case to the detected model settings allows the precision of the measurements to be controlled while the energy consumption per exposure is reduced.

[0114] This can be seen by observing the measurement results obtained with full illumination, i.e., in this case illumination according to the six considered modulation frequencies (Fig. 7a for the “thick” group and Fig. 8a for the “thin” group), or illumination limited to the optimal illumination case (Fig. 7b for the “thick” group and Fig. 8b for the “thin” group).

[0115] For the “thick” group, by the selection of only two of the six possible modulation frequencies (in this case 50 Hz and 200 Hz) and by the analysis of only the amplitudes (or equivalently, modules) of the signal components detected at each of these frequencies, The relative variation of the mean squared error is 7%. Consumption will be reduced by at least a third, The mean squared error was 10.5 mg / dL to 11.2 mg / dL, which was well below the target threshold of 20 mg / dL, which was considered the maximum acceptable RMSE in this case when selecting the variables of interest for the optimal irradiation case.

[0116] For the “thin layer” group, by the selection of only two of the six possible modulation frequencies (in this case 50 Hz and 400 Hz) and by the analysis of only the amplitudes (or equivalently, modules) of the signal components detected at each of these frequencies, The relative variation of the mean squared error is 20%. Consumption will be reduced by at least a third, The mean square error was 4.9mg / dL~5.8mg / dL, which was well below the target threshold of 20mg / dL, which is considered to be the maximum allowable RMSE in this case, when selecting the variable of interest for the optimal irradiation case. Therefore, even though the relative variation of the mean square error in this case cannot be ignored, the absolute value is kept under control when selecting the optimal irradiation case.

[0117] On the other hand, it can be seen that no technical effect is obtained when there is no association of model settings for a specific irradiation case and a specific inverse model, i.e. when a single inverse model is trained for the entire model settings of the two groups "thin layer" and "thick layer" and irradiation is carried out according to an irradiation case that applies to all model settings of these two groups "thin layer" and "thick layer".

[0118] The optimal irradiation case selected for the entire two groups is seen in Figure 6c, and the selected irradiation parameters are (wavelength 1034 cm -1 , two different modulation frequencies 200 Hz and 400 Hz) and an analysis of the module (equivalent to amplitude) of the signal detected at each of these frequencies is required.

[0119] As can be seen in Figures 9a and 9b, if one goes from illumination according to six different modulation frequencies (RMSE = 13.0 mg / dL) to illumination according to the two modulation frequencies identified as the most influential (i.e. 200 Hz and 400 Hz) (RMSE = 26.9 mg / dL) and analyzes only the amplitude of the signal detected at each of these frequencies, the mean square deviation is multiplied by three. The RMSE threshold, fixed in this example at 20 mg / dL, is also exceeded, indicating that other influencing variables must also be taken into account to adhere to this threshold.

[0120] The correspondence table may possibly include optimal irradiation cases for the whole group, especially in terms of initial irradiation. In the above example, it is understood that in order to meet the criteria regarding the RMSE value for initial irradiation, it would be necessary to include at least one additional modulation frequency, i.e. 50 Hz, and perhaps even reduce consumption, but only by a factor of two.

[0121] In the above, it can be seen in Figures 7a to 9b that the technical effect of reducing consumption with precision-controlled measurements is obtained, in particular by a correspondence table which makes it possible to segment the set of possible conditions of the target environment 2 into a number of categories and to associate each of these categories with an optimal irradiation case.

[0122] In this case, the number of exposure parameters (i.e., the two quantities that carry the most information) was chosen to keep the RMSE below 20 mg / dL even for exposures that do not include all possibilities, and the accuracy is better than this threshold for hypoglycemia, according to the field of blood glucose sensors. However, it is also possible to select three, four, or more of all possible exposure parameters to get the RMSE value below another threshold (e.g., 15 mg / dL, 10 mg / dL, or 5 mg / dL).

[0123] The number of illumination parameters selected results from a compromise between the power available for each illumination and the acceptable RMSE.

[0124] These pairs (CMk, Iopt.k) or (category of model setting k, Iopt.cat.k) are stored in a correspondence table in the memory of the non-invasive sensor 1 (eg, the memory of the simulation module 15 and / or the adaptation module 14).

[0125] Also stored in the memory of the non-invasive sensor 1 (eg, in the memory of the simulation module 15 and / or the adaptation module 14) is an inverse model corresponding to each model setting CMk or each model setting category.

[0126] At this stage, the elements necessary for carrying out the measurement method according to the present invention are ready.

[0127] It is already understood that by the correspondence table, it is not necessary to establish the complete optical spectrum of the target for the target conditions to be measured during the measurement. Automatic selection of the optimal illumination case corresponding to the current model settings, i.e. the values ​​of the Level 1 parameters that best represent the target environment 2 of the current measurement, can be performed by the correspondence table.

[0128] The measurement method according to the invention therefore allows to reduce the energy consumption of the sensor 1 compared to prior art processes with the same or even higher accuracy.

[0129] Methods for measuring parameters of interest A method for measuring a parameter of interest by a non-invasive sensor 1 based on indirect photoacoustic detection is shown in Figure 2. It comprises the following steps: a) Initialization Based on a predetermined criterion, an initial model setting of the target environment 2 for the first exposure is selected. For example, in the case of FIG. 2, the initial model setting CMirrad is the model setting CMk. There are various options available for this initialization. In particular, when measuring interstitial blood glucose from a database of model settings and / or a database of experimental measurements, a set of level 1 parameters and average blood glucose can be determined for a patient population or a given patient. This set of average parameters corresponds to the initial model setting C, which is most likely to be optimal for the next measurement in the absence of other information, particularly in the absence of measurement history. b) Using the correspondence table, the processor of the adaptation module 14 determines the exposure case Iopt.k that is associated with the initial model setting (eg, CMk). c) In a first illumination, a first illumination of the target environment 2 is performed by the light source 11a according to the illumination parameters of the optimal illumination case Iopt.k for the model setting CMk. d) PA detection An actual photoacoustic signal generated in response to the illumination is detected by a photoacoustic detection cell 12. A signal processing module 13 receives and processes the actual photoacoustic signal, if necessary, and transmits it to an adaptation module 14 after processing. e) Solving the inverse problem The processor of the adaptation module 14, which executes the learned inverse model (specifically corresponding to at least the model setting CMk), receives as input the actual photoacoustic signal and the irradiation case Iopt.k (as shown in FIG. 5) and determines the ongoing model setting of the target environment 2, taking into account the CMmes and the parameters of interest. f) Validation of model settings The adaptation module 14 compares the model setting CMmes of the ongoing target environment 2 with the model setting CMirrad used for the irradiation (model setting CMk from an initialization step for the first measurement (possibly different model setting CMl for the other measurements)). Two scenarios are possible: Case 1) If the model setting of the target environment 2 measured as CMmes is identical to the model setting CMirrad used for irradiation, then the selected irradiation parameters are optimal for the ongoing physiological situation (note that it is not known in advance and changes over time), as are the inverse models used to determine the parameters of interest. As a result, the parameters of interest determined in step f by the adaptation module 14 are the result of the measurement. Case 2) If the model setting of the target environment 2 measured as CMmes is different from the model setting CMirrad, the processor of the adaptation module uses the correspondence table to determine a new model setting CMl (l ≠ k) that is more suitable for the model setting CMmes estimated by the adaptation module 14. In particular, it is possible to select CMl = CMmes. However, other selections are also possible, especially if one wants to take into account the history of measurements over a long period of time, taking into account at least two model settings selected in the past. The processor of the adaptation module 14 then searches in the correspondence table for the optimal irradiation case Iopt.l for the model setting CMl of the target environment 2 and sends the corresponding parameters to the irradiation device 11. The target environment 2 is then subjected to a new illumination (subsequent illumination) according to the illumination parameters of the illumination case Iopt.l, and step d) "detection" and, if necessary, photoacoustic signal processing and step e) "solving the inverse problem" are repeated again.

[0130] The method may include at most one step f) "Verification of model settings."

[0131] Alternatively, the method may include repeating step f) "Validating the model configuration" again.

[0132] The model setting of the target environment 2, denoted as CMmes, may differ from the model setting CMirrad. According to the initialization criteria (CMirrad is the "average" model setting, especially in the absence of a patient measurement history (for the first exposure)), or Based on past measurements (CMirrad is the best model for patients with known past post-irradiation measurements), CMirrad has been selected.

[0133] The validation of CMirrad is therefore performed by means of additional information obtained by the ongoing exposure, i.e. the signal detected by the photoacoustic detection cell 12. If by chance CMmes=CMirrad, then no new exposures need to be performed and the value of the parameter of interest is the most accurate value that can be obtained. On the other hand, the validation step of the model setup gives additional information, i.e. a confirmation that the accuracy of the measurements in this case is maximal.

[0134] If CMmes differs from CMirrad, adaptation of the irradiation parameters and model settings will improve the measurement accuracy with at least one additional irradiation, but energy consumption is always kept under control and a second validation step ensures that the measurement is as accurate as possible.

[0135] If we allow the verification step to be repeated again (as shown in FIG. 2), we obtain at the output of the process additional information that the measurement accuracy is at its maximum.

[0136] Typically in this case, measurements of the parameters of interest are obtained after the first or second exposure, although limitations on the convergence of the process and / or on the energy consumption can be expected to limit the number of validation steps for the model setup.

[0137] In all cases, it turns out that the selection of the irradiation case Ij based on the correspondence table makes it possible to limit the number of modulation frequencies and wavelengths used for irradiation, by retaining only those values ​​that have non-redundant information about the ongoing model setting of the physiological target environment 2, and is sufficient to obtain the desired and / or optimal measurement accuracy to limit the energy consumption of the sensor to a given value.

[0138] For example, if the physiological situation corresponds to three layers, each characterized by two concentrations (e.g., water and blood glucose), then the solution of the inverse problem is a problem with 2 × 3 = 6 unknowns. By knowing the amplitude and phase of each of the three components of the photoacoustic signal corresponding to the modulation frequency and appropriately chosen wavelengths, the inverse problem should be solvable with the desired accuracy. The difficulty remains in the optimal selection of these modulation frequencies and optimal wavelengths, which is solved using a correspondence table.

[0139] Thus, the consumption of the sensor 1 is limited or controlled.

[0140] Even if two or three exposures are required, each exposure step requires only limited power, which is less than the power required to obtain a complete absorption spectrum for each possible frequency modulation.

[0141] This model setup validation step allows: an automatic recalibration of the sensor 1 so that the inverse model allowing the estimation of the value of the parameter of interest of the detected optoacoustic signal without prior knowledge of the current physiological situation is the inverse model most relevant to the current physiological situation; Adjustment of the irradiation case so that the parameters used for the subsequent irradiations are optimized with respect to the information about the target environment, in particular the information provided by the last detected photoacoustic signal; becomes possible.

[0142] To further optimize the measurement process according to the present invention, several embodiments are possible.

[0143] In particular, the artificial intelligence model of the simulation device 15 can be pre-trained on a database of simulated model settings and then re-trained on a database of empirical model settings corresponding to real situations of a given patient or a given group of patients, thereby enabling the inverse model to correctly predict real situations without requiring an exhaustive set of experimental data. In this case, the inverse learning model can include a transfer learning phase ("transfer learning"). [Explanation of symbols]

[0144] 1. Non-invasive Sensor Based on Indirect Photoacoustic Detection 11 Irradiation Device 11a light source 11b Device for modulating the intensity of light source 11a 11c A control device for the modulation frequency fmod of the intensity of the light source 11a 12 Detection Cell 13 Signal Processing Module 14 Adaptation Module 15 Simulation Module

Claims

1. A method for measuring parameters of interest in a target environment (2) using a non-invasive sensor (1) based on photoacoustic detection or photothermal detection, a) Prepare a sensor, Light source (11a), A device for controlling the irradiation parameters of the light source (11a), A detection cell (12) configured to detect acoustic or thermal signals, A memory that stores a correspondence table including model settings (CMk) representing the given states of the target environment (2) and optimal irradiation cases (Iopt.k) that include sets of irradiation parameters, wherein each model setting (CMk) is associated with an optimal irradiation case. An adaptive module (14) that exchanges information with the detection cell (12) and the light source irradiation parameter control device, comprising a processor configured to receive an irradiation case (Ij) including a set of irradiation parameters and an acoustic or thermal signal as inputs, and to execute an inverse modeling algorithm that provides a model setting (CMk) and the values ​​of the parameters of interest as outputs, and To provide a sensor equipped with, b) The adaptive module (14) selects the initial irradiation model setting (CMirrad), c) The adaptive module (14) determines from the correspondence table the optimal irradiation case for the selected irradiation model setting (CMirrad), i.e., an irradiation case that enables the measurement of the parameter of interest with a measurement data amount that allows for a predetermined accuracy and / or minimum energy consumption. d) The light source (11a) irradiates the target environment (2) according to the set of irradiation parameters of the optimal irradiation case, e) The detection cell (12) detects an acoustic or thermal signal generated in response to irradiation, f) The processor (14) of the adaptive module executes the inverse modeling algorithm, receives the acoustic or thermal signal detected by the detection cell (12) and the optimal irradiation case used for irradiation as input, and returns the current model settings (CMmes) and the estimated values ​​of the parameters of interest (Pest) as output, g) The processor of the adaptive module (14) evaluates the selected irradiation model setting (CMirrad) by comparison with the current model setting (CMmes), and only if the selected irradiation model setting (CMirrad) is different from the current model setting (CMmes), g1) the adaptive module (14) receives the current model setting (CMmes) as input and returns a new irradiation model setting (CMirrad) as output, after which steps c), d), e), and f) are repeated. h) The measured value of the parameter of interest (Pmes) obtained by the sensor is the final estimated value (Pest) of the parameter of interest. Measurement methods, including those mentioned above.

2. The measurement method according to claim 1, further comprising repeating g) after f).

3. In advance, I. The measurement method according to claim 1, comprising generating a correspondence table from a processor and a database of model settings including multiple terms (model setting (CMk), irradiation case (Ij), parameters of interest) and acoustic or thermal signals detected by the detection cell (12) associated with each multiple term, and storing the correspondence table in the memory of the non-invasive sensor (1).

4. II. The measurement method according to claim 3, comprising the processor learning at least one inverse modeling algorithm from the database of model settings and storing it in the memory of the non-invasive sensor (1).

5. The measurement method according to claim 3, wherein at least a portion of the acoustic or thermal signal detected by the detection cell (12) associated with the multiple terms stored in the model setting database is simulated, i.e., generated, by a computerized simulation device.

6. A non-invasive sensor (1) based on photoacoustic or photothermal detection is configured to measure parameters of interest in a target environment (2), Light source (11a), A device for controlling the irradiation parameters of the light source (11a), A detection cell (12) configured to detect acoustic or thermal signals, A memory that stores a correspondence table including model settings (CMk) representing a given state of the target environment (2) and optimal irradiation cases (Iopt.k) that include sets of irradiation parameters, wherein each model setting (CMk) is associated with an optimal irradiation case (Iopt.k), and Equipped with, The non-invasive sensor is an adaptive module (14) configured to exchange information with the light source for controlling the irradiation parameters of the detection cell (12) and the light source (11a), further comprising a processor configured to run an inverse modeling algorithm that receives an irradiation case (Ij) including a set of irradiation parameters and an acoustic or thermal signal as inputs and provides a model setting (CMk) and values ​​of the parameters of interest as outputs, The adaptive module (14) i. Select the initial irradiation model setting (CMirrad), ii. Determine the optimal irradiation case corresponding to the irradiation model setting from the aforementioned correspondence table, iii. Transmitting the optimal irradiation case to the light source irradiation parameter control device, iv. Receiving the signal detected by the detection cell (12), v. Determining the current model settings (CMmes) and estimated values ​​(Pest) of the parameters of interest based on the received detected photoacoustic or photothermal signals and the optimal irradiation case, vi. Evaluate the irradiation model setting (CMirrad) by comparing it with the current model setting (CMmes), vii. Only when the irradiation model setting (CMirrad) of the selected target layered environment differs from the model setting (CMmes) of the current target layered environment to be compared, a new irradiation model setting (CMirrad) for receiving the current model setting (CMmes) is determined, a new optimal irradiation case corresponding to the new irradiation model setting (CMirrad) is determined from the lookup table, and the new optimal irradiation case is transmitted to the irradiation parameter control device, thereby causing the light source to irradiate the target layered environment according to the set of irradiation parameters of the new optimal irradiation case, the detection cell to detect a new thermal or acoustic signal generated in response to the irradiation, and transmit it to the processor of the adaptive module (14) which is configured to repeat iv, v, vi, and vii again. Based on the final estimated value (Pest) of the parameter of interest, the measured value (Pmes) of the parameter of interest is determined. A non-invasive sensor (1) is further configured to perform the following.

7. A computer program comprising instructions to cause the non-invasive sensor (1) according to claim 6 to perform steps of a method for measuring parameters of interest in a target environment by a non-invasive sensor based on photoacoustic detection or photothermal detection, wherein the method is a) Prepare a sensor, Light source and A device for controlling the irradiation parameters of the aforementioned light source, A detection cell configured to detect acoustic or thermal signals, A memory that stores a correspondence table including an optimal irradiation case which includes a model setting representing a given state of the target environment and a set of irradiation parameters, and each model setting is associated with an optimal irradiation case, and An adaptive module that exchanges information with the detection cell and the light source irradiation parameter control device, comprising a processor configured to receive an irradiation case including a set of irradiation parameters and an acoustic or thermal signal as inputs, and to execute an inverse modeling algorithm that provides model settings and values ​​of parameters of interest as outputs. To provide a sensor equipped with, b) The adaptive module selects the initial irradiation model setting, c) The adaptive module determines from the correspondence table the optimal irradiation case for the selected irradiation model setting, i.e., an irradiation case that enables the measurement of the parameter of interest with a quantity of measurement data that allows for a predetermined accuracy and / or minimum energy consumption, d) The light source irradiates the target environment according to the set of irradiation parameters of the optimal irradiation case, e) The detection cell detects an acoustic or thermal signal generated in response to irradiation, f) The processor of the adaptive module executes the inverse modeling algorithm, receives the acoustic or thermal signal detected by the detection cell and the optimal irradiation case used for irradiation as input, and returns the current model settings and estimates of the parameters of interest as output. A computer program that includes [this].

Images