A method and system for intelligently identifying anesthetic gas components

Abstract

The present invention relates to the technical field of spectral data processing, and in particular to a method and system for intelligently identifying anesthetic gas components. First, spectral data and corresponding flow rate data of multiple anesthetic gas mixture samples are obtained; then, based on the spectral data, a content contribution model of a single anesthetic gas is constructed; then, based on the flow rate data, the content contribution model of the single anesthetic gas is modified to obtain a revised content contribution model; then, based on the revised content contribution model, an optimization problem based on a regression model is constructed; finally, the objective function of the optimization problem is solved to obtain an anesthetic gas estimation model. The present invention utilizes spectral data at different flow rates and original anesthetic gas concentration ratios to further quantify the contribution of different wavebands to the content of each anesthetic gas substance, so that the contribution at different flow rates can be adaptively adjusted to improve the detection accuracy of the detection system for anesthetic gases at different flow rates.

Description

Technical Field

[0001] The present invention relates to the technical field of spectral data processing, and in particular to an intelligent identification method and system for anesthetic gas components. Background Art

[0002] During surgery, it is necessary to identify the components of each anesthetic gas in the mixed anesthetic gas to ensure that the patient is in an appropriate anesthesia state. Usually, the content of multiple substances in the mixed anesthetic gas is identified based on regression.

[0003] In the prior art, there is a technical problem of insufficient accuracy in identifying the contents of multiple substances in mixed anesthetic gases. Summary of the Invention

[0004] In order to solve the technical problem of inaccurately identifying the contents of multiple substances in mixed anesthetic gases, the present invention aims to provide an intelligent method and system for identifying anesthetic gas components. The technical solutions adopted are as follows:

[0005] An intelligent method for identifying anesthetic gas components, the method comprising:

[0006] Acquiring spectral data and corresponding flow rate data of a plurality of anesthetic gas mixture samples, wherein different anesthetic gas mixture samples have different flow rates;

[0007] Based on the spectral data, constructing a content contribution model of a single anesthetic gas;

[0008] Based on the flow rate data, the content contribution model of the single anesthetic gas is modified to obtain a revised content contribution model; wherein the revised content contribution model takes into account the influence of the flow rate data on the spectral data; wherein the influence is used to indicate that when the flow rate data is higher than a certain threshold, the peak shape of the spectral data becomes narrower, and when the flow rate data is lower than a certain threshold, the peak shape of the spectral data becomes wider;

[0009] Based on the modified content contribution model, an optimization problem based on a regression model is constructed; wherein the objective function of the optimization problem is the difference between the actual content of a certain anesthetic gas detected and the estimated content calculated by the regression model;

[0010] An objective function of the optimization problem is solved to obtain an anesthetic gas estimation model, wherein the anesthetic gas estimation model is used to perform intelligent identification of anesthetic gas.

[0011] Furthermore, the step of constructing a content contribution model of a single anesthetic gas based on the spectral data includes:

[0012] Determining characteristic wavelength band intervals of each anesthetic gas in the anesthetic mixed gas sample based on the spectral data;

[0013] A content contribution model of different bands to a single anesthetic gas within the characteristic band interval is constructed; wherein the content contribution model takes into account the response estimation value of a single anesthetic gas in different bands within the characteristic band interval.

[0014] Furthermore, the step of modifying the content contribution model of the single anesthetic gas based on the flow rate data to obtain a revised content contribution model includes:

[0015] Determining an adaptive gamma value of the spectral data of the anesthetic gas mixture sample based on the flow rate data; wherein the adaptive gamma value is used to represent the difference between the skewness coefficient of the absorbance at the maximum flow rate and the skewness coefficient of the absorbance in each band within each characteristic band interval when the flow rate data is carried, wherein the difference is mapped by an exponential function;

[0016] Based on the adaptive Gamma value, the content contribution model is corrected to obtain a corrected content contribution model.

[0017] Furthermore, the characteristic band interval is determined by the following method:

[0018] For the spectral data of each anesthetic gas, determining the wavelength band corresponding to the maximum absorbance of the spectral data;

[0019] Starting from the wavelength corresponding to the maximum absorbance of the spectral data, a sliding window of a preset length is slid to both sides by a distance of a number of sliding windows;

[0020] The wavelength band within the sliding window and the wavelength band corresponding to the maximum value of the spectral data absorption rate are determined as the characteristic wavelength band interval.

[0021] Furthermore, the expression of the content contribution model is: the content contribution model is expressed as the product of the proportional coefficient of the estimated absorption rate of any anesthetic gas in a certain band to the estimated absorption rates of other anesthetic gases and the spectral absorption rate of the corresponding anesthetic gas in the band, wherein the product is subjected to maximum and minimum normalization processing.

[0022] Furthermore, in the objective function of the optimization problem, the actual content of the anesthetic gas actually detected is determined by the following method:

[0023] The actual content of a certain anesthetic gas actually detected is represented by a preset training sample; wherein the training sample includes the component content of at least one anesthetic gas substance of at least one anesthetic mixed gas sample under at least one gas flow rate.

[0024] Furthermore, in the expression of the objective function of the optimization problem, the estimated content calculated by the regression model is determined by the following method:

[0025] Construct the coefficient vector to be solved;

[0026] The coefficient vector to be solved is multiplied by the correction coefficient matrix and the spectral data absorbance vector to obtain the product value; wherein the correction coefficient matrix is ​​the component content of any anesthetic gas substance in any anesthetic gas mixture sample, when the anesthetic gas flow rate data is a certain value Maintenance positive coefficient matrix; where, is the number of bands in the characteristic band interval, and the spectral data absorption rate vector is the spectral data absorption rate vector of the corresponding anesthetic gas substance in the characteristic band interval;

[0027] The multiplied value is added to the bias parameter to be solved, and then at least the number of anesthetic mixed gas samples and the anesthetic gas flow rate data are summed to obtain the estimated content calculated by the regression model.

[0028] Furthermore, the objective function of the optimization problem is solved by using Lagrange number multiplication or gradient descent method.

[0029] Furthermore, the expression of the anesthetic gas estimation model is:

[0030] The product of the solved coefficient vector, component content, and spectral absorptivity vector is added with the solved bias parameter; wherein the component content is the component content of the target anesthetic gas substance in the target anesthetic gas mixture sample; wherein the spectral absorptivity vector represents the spectral absorptivity vector corresponding to multiple bands within the characteristic band interval of the target anesthetic gas substance in the target anesthetic gas mixture sample.

[0031] An intelligent anesthetic gas composition identification system, comprising:

[0032] An acquisition module, which is used to acquire spectral data and corresponding flow rate data of multiple anesthetic gas mixture samples; wherein different anesthetic gas mixture samples have different flow rates;

[0033] A first building module is configured to build a content contribution model of a single anesthetic gas based on the spectral data;

[0034] a modification module for modifying the content contribution model of the single anesthetic gas based on the flow rate data to obtain a modified content contribution model; wherein the modified content contribution model takes into account the influence of the flow rate data on the spectral data; wherein the influence is used to indicate that when the flow rate data is higher than a certain threshold, the peak shape of the spectral data becomes narrower, and when the flow rate data is lower than a certain threshold, the peak shape of the spectral data becomes wider;

[0035] A second construction module is configured to construct an optimization problem based on a regression model based on the modified content contribution model; wherein the objective function of the optimization problem is the difference between the actual content of a certain anesthetic gas detected and the estimated content calculated by the regression model;

[0036] A solution module is used to solve the objective function of the optimization problem, thereby obtaining an anesthetic gas estimation model, wherein the anesthetic gas estimation model is used to perform intelligent identification of anesthetic gas.

[0037] The present invention has the following beneficial effects:

[0038] The present invention first obtains spectral data and corresponding flow rate data of multiple anesthetic gas mixture samples; wherein different anesthetic gas mixture samples have different flow rates; then, based on the spectral data, constructs a content contribution model of a single anesthetic gas; then, based on the flow rate data, modifies the content contribution model of the single anesthetic gas to obtain a modified content contribution model; wherein, the modified content contribution model takes into account the influence of the flow rate data on the spectral data; wherein, the influence is used to indicate that when the flow rate data is higher than a certain threshold, the peak shape of the spectral data becomes narrower, and when the flow rate data is lower than a certain threshold, the peak shape of the spectral data becomes wider. width; then, based on the modified content contribution model, an optimization problem based on a regression model is constructed; wherein, the objective function of the optimization problem is the difference between the actual content of a certain anesthetic gas actually detected and the estimated content calculated by the regression model; finally, the objective function of the optimization problem is solved to obtain an anesthetic gas estimation model. This method uses spectral data under different flow rates and original anesthetic gas concentration ratios to further quantify the contribution of different bands to the content of each anesthetic gas substance, so that the contribution at different flow rates can be adaptively adjusted to improve the detection accuracy of the detection system for anesthetic gases at different flow rates. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] In order to more clearly illustrate the technical solutions and advantages of the embodiments of the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art descriptions. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0040] Figure 1 A flowchart of an intelligent identification method for anesthetic gas components provided by one embodiment of the present invention;

[0041] Figure 2 A flowchart of step S120 provided in one embodiment of the present invention;

[0042] Figure 3 A flowchart of step S130 provided in one embodiment of the present invention;

[0043] Figure 4 This is a block diagram of an intelligent anesthetic gas composition identification system provided by one embodiment of the present invention. DETAILED DESCRIPTION

[0044] To further illustrate the technical means and effectiveness of the present invention in achieving its intended objectives, the following, in conjunction with the accompanying drawings and preferred embodiments, describes in detail the specific implementation, structure, features, and effectiveness of an intelligent anesthetic gas composition identification method and system according to the present invention. In the following description, references to "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics of one or more embodiments may be combined in any suitable manner.

[0045] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0046] The specific scheme of the intelligent identification method and system for anesthetic gas components provided by the present invention will be described in detail below with reference to the accompanying drawings.

[0047] Among related technologies, spectral analysis is a technique that determines the composition and properties of a substance by measuring its interaction with electromagnetic radiation. In spectral analysis, the interaction between a substance and electromagnetic radiation (such as visible light, infrared, and ultraviolet light) produces specific spectral signatures that can be used to identify information such as the substance's composition, concentration, and structure. Linear regression is a statistical modeling technique used to analyze the linear relationship between independent and dependent variables. In linear regression, a linear relationship between the dependent and independent variables is assumed, and a linear equation is fitted to describe the relationship.

[0048] During surgery, ensuring the patient is in an appropriate state of anesthesia is crucial. Excessively high or low anesthesia can lead to adverse consequences, including issues with recovery of consciousness during surgery, pain perception, movement during surgery, and postoperative recovery time. In anesthetic gas monitoring, gas flow rates can vary due to different breathing patterns or equipment operating conditions. In spectral analysis, these changes in flow rate cause changes in optical path length or spectral intensity, affecting spectral characteristics. These changes challenge the accuracy of spectral analysis at varying flow rates. When the gas flow rate changes, the speed of light propagating through the gas also changes, resulting in a change in optical path length. This change in optical path length affects the path length for light to interact with matter, which in turn affects spectral characteristics. Furthermore, changes in flow rate can affect spectral intensity, or the intensity or energy of light. In spectral analysis, spectral intensity is a parameter used to represent the intensity or energy of light at different wavelengths. Changes in flow rate can cause changes in spectral intensity. This change affects spectral characteristics, resulting in different spectral characteristics when analyzing at different flow rates.

[0049] Existing methods for identifying the content of multiple substances in anesthetic gases are usually based on regression. These methods use response values ​​in different bands to estimate the concentration of anesthetic gases. However, as the flow rate changes, the contribution of response values ​​in different bands to the concentration of anesthetic gases will change, resulting in the accuracy of existing linear regression methods at different flow rates being affected, leading to large errors in the detection of different anesthetic substance components at higher flow rates. Therefore, it is considered to use spectral data at different flow rates and original anesthetic solution concentration ratios to further quantify the contribution of different bands to the content of each substance, so that the contribution at different flow rates can be adaptively adjusted to improve the detection accuracy of the detection system for anesthetic gases at different flow rates.

[0050] Therefore, in the prior art, there is a technical problem that the content of multiple substances in the mixed anesthetic gas is not accurately identified. In response to this technical problem, the present invention first obtains the spectral data and corresponding flow rate data of multiple anesthetic mixed gas samples; wherein, different anesthetic mixed gas samples have different flow rates; then, based on the spectral data, a content contribution model of a single anesthetic gas is constructed; then, based on the flow rate data, the content contribution model of the single anesthetic gas is modified to obtain a revised content contribution model; wherein, the revised content contribution model takes into account the influence of the flow rate data on the spectral data; wherein, the influence is used to indicate that when the flow rate data is higher than a certain threshold, the peak shape of its spectral data becomes narrower, and when the flow rate data is lower than a certain threshold, its spectral data becomes narrower. peak broadening; then, based on the modified content contribution model, an optimization problem based on a regression model is constructed; wherein the objective function of the optimization problem is the difference between the actual content of a certain anesthetic gas actually detected and the estimated content calculated by the regression model; finally, the objective function of the optimization problem is solved to obtain an anesthetic gas estimation model. This method uses spectral data at different flow rates and original anesthetic gas concentration ratios to further quantify the contribution of different bands to the content of each anesthetic gas substance, so that the contribution at different flow rates can be adaptively adjusted to improve the detection accuracy of the detection system for anesthetic gases at different flow rates.

[0051] Example 1:

[0052] See also Figure 1 , which shows a method flow chart of an intelligent identification method for anesthetic gas components provided by one embodiment of the present invention, the method comprising:

[0053] Step S110: Acquire spectral data and corresponding flow rate data of a plurality of anesthetic gas mixture samples; wherein different anesthetic gas mixture samples have different flow rates.

[0054] In this embodiment, the aforementioned multiple anesthetic gas mixture samples may be anesthetic gas mixture samples at multiple moments. It is understandable that the flow rates of the anesthetic gas mixture samples at different moments are different.

[0055] In this embodiment, the above-mentioned anesthetic gas mixture sample may be a mixture of multiple anesthetic gases, for example, a mixture of two or more anesthetic gases selected from desflurane, isoflurane, enflurane, sevoflurane and halothane.

[0056] In one possible example scenario, the following structure of an anesthesia ventilator can be provided: an anesthetic solution mixture container, an oxygen solution container, an anesthetic solution mixture flowmeter, an oxygen solution flowmeter, a vaporizer, a rapid oxygen supply switch, a carbon dioxide absorber, a safety valve, a carbon dioxide analyzer, a ventilation meter, a ventilator, an air storage bag, and an exhaust port. The anesthetic solution mixture container and the oxygen solution container are connected to the vaporizer input via a pipe, and an anesthetic solution mixture flowmeter is provided between the anesthetic solution mixture container and the vaporizer input, and an oxygen solution flowmeter is provided between the oxygen solution container and the vaporizer input. The vaporizer output is connected to the carbon dioxide absorber input, the inlet and outlet ports of the ventilator and air storage bag, and the ventilation meter input via pipes. A safety valve is also provided at the intersection of the vaporizer output, the inlet and outlet ports of the ventilator and air storage bag, and the ventilation meter input. The carbon dioxide absorber output is connected to the output port via the carbon dioxide analyzer, and the ventilation meter is also connected to the output port via a pipe. The output port is used to deliver gas to the patient. In this example scenario, during the operation of an anesthesia ventilator, real-time spectral data of the vaporizer's output gas is acquired to measure the content of each anesthetic substance in the gas at different times to meet changing surgical needs and ensure the patient is properly anesthetized. The vaporizer contains a mixed anesthetic solution, which is delivered to the breathing circuit along with carrier gases such as oxygen. To accurately identify the concentration of each substance in the output gas, the original substance content estimation method is improved by considering the correlation between the output gas flow rate and the changes in the spectral data.

[0057] Therefore, in this possible scenario example, the above-mentioned acquisition action can be to obtain infrared spectrum data and corresponding flow rate data of multiple anesthetic gas mixture samples output by the vaporizer in the anesthesia ventilator. It is understandable that the anesthetic gas mixture samples can be at different times, and the flow rates of the anesthetic gas mixture samples at different times are different. It is understandable that the cost content of the above-mentioned multiple anesthetic gas mixture samples should be the same, but the flow rates are different. In the infrared spectrum data, there are differences in the spectral curves of different anesthetic gases, the wavelength range is 700nm-1400nm, the wavelength resolution is 1nm, and there are many overlapping areas between the absorption peaks of different anesthetic gases, for example, there are many overlapping areas between the absorption peaks of desflurane, isoflurane, enflurane, and sevoflurane.

[0058] Step S120: constructing a content contribution model of a single anesthetic gas based on the spectral data.

[0059] In some embodiments, as Figure 2 As shown, the above step S120 may include the following sub-steps:

[0060] Step S211: determining the characteristic wavelength band interval of each anesthetic gas in the anesthetic mixed gas sample based on the external spectrum data.

[0061] It's understandable that the flow rate of anesthetic gas varies during surgery. In spectral analysis, changes in gas flow rate can lead to changes in the optical path length. When the gas flow rate increases, the speed at which gas molecules pass through the detection area increases, shortening the optical path length and weakening the intensity of the absorption peak. Conversely, when the gas flow rate decreases, the optical path length increases, increasing the intensity of the absorption peak. This change in path length directly affects the intensity of the absorption peak in the spectral data. When the intensity and shape of the absorption peak in the spectral data change, the characteristics of the data also change. Traditional multivariate linear regression models are based on linear relationships between features. If the characteristics of the data change, the model's assumptions no longer hold, resulting in a decrease in model accuracy. Therefore, when the intensity and shape of the absorption peak in the spectral data change, traditional regression models cannot accurately capture the relationship between the data, resulting in a decrease in accuracy.

[0062] In this embodiment, the characteristic band interval of each anesthetic gas in the anesthetic mixed gas sample can be determined by the following method. Specifically, first, for the spectral data of each anesthetic gas, the band corresponding to the maximum value of its spectral data absorption rate is determined; then, with the band corresponding to the maximum value of the spectral data absorption rate as the starting point, a sliding window of a preset length is slid to both sides by the distance of several sliding windows, and finally the band within the sliding window and the band corresponding to the maximum value of the spectral data absorption rate are determined as the characteristic band interval.

[0063] Specifically, in some embodiments, the spectral data of a single anesthetic gas in the above anesthetic gas mixture sample can be defined as , for income , select the band corresponding to the maximum value of its spectral data absorption rate and record it as its characteristic band However, due to the influence of band offset, the band may have a certain offset in different measurement environments. Therefore, with this band as the center, multiple bands are selected from the two sides of the neighborhood as the characteristic band interval corresponding to the anesthetic gas until the average internal absorption rate of the sliding window of each band in the neighborhood is less than ,in Indicates the spectral data of anesthetic gases, Indicates the characteristic band corresponding to the maximum absorbance of the spectral data. It can be expressed as The maximum value of the absorption rate of the spectrum data of the anesthetic gas. 0.6 is a specific coefficient, and in some embodiments it can also be 0.7 or 0.5. When the mean absorption rate of the sliding window of each band in the neighborhood is less than Stop the range expansion.

[0064] In some embodiments, the selected sliding window radius may be 20 nm, so as to obtain the characteristic waveband interval corresponding to each substance.

[0065] Step S212: constructing a content contribution model of different bands within the characteristic band interval to a single anesthetic gas; wherein the content contribution model takes into account the response estimation value of a single anesthetic gas in different bands within the characteristic band interval.

[0066] It is understandable that the spectral data of the anesthetic gas mixture sample contains multiple different material components, such as desflurane, isoflurane, enflurane, and sevoflurane. There are many overlapping bands between the characteristic bands corresponding to each of them. During the regression analysis, if the same weight coefficient is assigned to different bands, some bands will be affected by the band overlap, resulting in deviations in the contribution of spectral data at different concentration ratios to the content of individual substances. Therefore, the weight coefficients of each material component in the spectral data of the anesthetic gas mixture in different bands are improved based on the changes in the spectral data corresponding to each substance.

[0067] In this embodiment, the content contribution model may be: (1)

[0068] In this formula (1), Expressed as In the spectrum data corresponding to the anesthetic gas mixture sample, Within the characteristic wavelength range of each substance, The contribution of each band to the content of the anesthetic gas is Indicates the The first bands, Indicates the The spectral data corresponding to the anesthetic gas mixture sample is in the band The spectral absorption rate under Indicates in band Next, there is The characteristic band interval of the anesthetic gas will pass through the band , The function is the maximum and minimum normalization, Indicates the The spectral data corresponding to the anesthetic gas mixture sample Anesthetic gases are present in the spectral data in the band The estimated value of the absorption rate under Indicates in band Next, The estimated absorption rate of the first anesthetic gas in this band accounts for the proportion coefficient of the average absorption rate estimated value of all anesthetic gases. If the band is less affected by multiple substances, the absorption rate of this band only represents the absorption rate of the first anesthetic gas. The higher the ratio is, the greater the contribution of this band to the content of a single substance will be. This is because the absorption rate of some mixed gas spectral data in the characteristic band range exists in the band with a small absorption rate. Due to its small absorption rate, the above-mentioned proportional coefficient will be more easily affected by noise, resulting in a large deviation in the contribution calculation result. Therefore, its absorption rate is used for further improvement.

[0069] , its specific expression is:

[0070] (2)

[0071] In this formula (2), Indicates the The characteristic band corresponding to the absorption peak of the anesthetic gas; Indicates the Samples in the band The spectral absorption rate under Indicates a single Spectral data of anesthetic gases in the band The absorption rate under Expressed as a single The spectrum data of the substance is in the band The absorption rate of this band is Expressed as The first bands, It represents the ratio of the absorbance in two different bands in the spectrum data of a single anesthetic gas substance. Indicates the optimal absorption band of a single anesthetic gas substance Regarding the absorbance in the spectral data of the anesthetic gas mixture sample, since the content of a single anesthetic gas substance is less affected by the other bands in its optimal absorption band, the reliability of estimating its response value in the other bands using the absorbance in this band is relatively high.

[0072] Step S130: Based on the flow rate data, the content contribution model of the single anesthetic gas is modified to obtain a revised content contribution model; wherein the revised content contribution model takes into account the influence of the flow rate data on the spectral data; wherein the influence is used to indicate that when the flow rate data is higher than a certain threshold, the peak shape of the spectral data becomes narrower, and when the flow rate data is lower than a certain threshold, the peak shape of the spectral data becomes wider.

[0073] In some embodiments, as Figure 3 As shown, the above step S130 may include the following sub-steps:

[0074] Step S311: Determine an adaptive gamma value of the spectral data of the anesthetic gas mixture sample based on the flow rate data; wherein the adaptive gamma value is used to characterize the difference between the skewness coefficient of the absorbance at the maximum flow rate and the skewness coefficient of the absorbance in each band within each characteristic band interval when the flow rate data is carried.

[0075] It is understandable that different airflow velocities affect spectral data. Specifically, in step S212, the contribution of each anesthetic gas within the corresponding characteristic band is analyzed only for the anesthetic gas spectral data corresponding to a single anesthetic solution sample. The calculation method for spectral data at different flow rates is the same. However, in practice, the carrier gas flow rate directly affects the sample's residence time and distribution in the detector. A higher carrier gas flow rate accelerates the sample's passage through the spectrometer, potentially narrowing the peak and thus improving analytical sensitivity. A lower carrier gas flow rate, while more evenly dispersing the sample in the detector, also prolongs the sample's detection time, leading to a broadened peak and affecting the absorbance distribution within the corresponding absorption peak's neighborhood.

[0076] Therefore, in some embodiments, it may be considered to adopt a gamma transformation approach to correct the distribution of the original contribution by combining the skewness coefficient of the distribution of the spectral data absorbance at different flow rates within the corresponding characteristic band interval.

[0077] In some implementations, the expression for the adaptive Gamma value may be:

[0078] (3)

[0079] In formula (3), Indicates the The flow rate data corresponding to the anesthetic gas mixture sample is The spectral data under, and its corresponding adaptive Gamma value, represents an exponential function with a natural constant as base, Indicates the The anesthetic gas mixture samples were In the case of The skewness coefficient of multiple absorption rates within the characteristic band interval of an anesthetic component. The multiple absorption rates within the characteristic band interval can be understood as the absorption rates corresponding to different bands within the characteristic band interval. Expressed as The anesthetic solution samples were In the case of The skewness coefficients of multiple absorption rates within the characteristic band interval of each anesthetic component.

[0080] Specifically, Indicates that the corresponding spectrum data at the current flow rate is The difference between the skewness coefficient of multiple absorbances within the characteristic band interval and the skewness coefficient at the maximum flow rate after mapping through the exponential function; when the flow rate data is small, the absorption peak area of ​​its spectral data within the corresponding characteristic band interval will expand. At this time, the absorbance of multiple bands that were originally small is small, and the absorbance will be larger at a smaller flow rate, which will cause the distribution of the absorbance to be more distributed on the left, resulting in a smaller skewness coefficient of the distribution, making the original content contribution corresponding to each substance larger, and its contribution under each band will be closer to 1, then the gamma transformation under each band will be given a larger gamma value, which will cause the original content contribution to be stretched to a greater extent within the range close to 1, and compress the interval range of the original low content contribution, so that the obtained content contribution is closer to the actual situation.

[0081] Step S312: Based on the adaptive Gamma value, the content contribution model is corrected to obtain a corrected content contribution model; wherein the corrected content contribution model takes into account the influence of the flow rate data on the spectral data.

[0082] In some embodiments, the impact of the above-mentioned flow rate data on the spectral data can be: when the flow rate data is higher than a certain threshold, the speed of the sample passing through the spectrometer will be accelerated, which will cause the peak shape to narrow, thereby improving the sensitivity of the analysis; when the flow rate data is lower than a certain threshold, although the sample is more evenly dispersed in the detector, it will also extend the detection time of the sample, resulting in peak broadening, affecting the absorption rate distribution in the neighborhood corresponding to the absorption peak.

[0083] In some embodiments, the specific expression of the above-mentioned modified content contribution model can be:

[0084] (4)

[0085] In formula (4), Represents the number of different anesthetic gas mixture samples obtained, Indicates the The corresponding flow rate of the anesthetic gas mixture sample is In the spectral data below, Band pair The contribution of the content of the substance, Indicates the The corresponding flow rate of the anesthetic solution sample is The spectral data under , the corresponding adaptive Gamma value, the expression of the adaptive Gamma value can be shown as formula (3), Expressed as the flow rate In the case of Band pair The contribution of the content of the substance.

[0086] The contribution of each of the above-mentioned anesthetic gas substances in the corresponding characteristic band interval is only analyzed for the anesthetic gas spectral data corresponding to a single anesthetic solution sample. The calculation method of the spectral data at different flow rates is the same, but in actual practice, the size of the carrier gas flow rate directly affects the residence time and distribution of the sample in the detector; when the carrier gas flow rate is large, the speed of the sample through the spectrometer will be accelerated, which may cause the peak shape to narrow, thereby improving the sensitivity of the analysis. When the carrier gas flow rate is small, although the sample is more evenly dispersed in the detector, it will also extend the detection time of the sample, resulting in peak broadening, affecting the distribution of the absorbance in the neighborhood corresponding to the absorption peak; therefore, it is considered to use the gamma transformation method to correct the distribution of the original contribution by combining the skewness coefficient of the distribution of the absorbance of the spectral data at different flow rates in the corresponding characteristic band interval.

[0087] Step S140: constructing an optimization problem based on a regression model based on the modified content contribution model; wherein the objective function of the optimization problem is the difference between the actual content of a certain anesthetic gas actually detected and the estimated content calculated by the regression model.

[0088] In this embodiment, the modified content contribution model obtained in the above steps can be used, that is, the improved content contribution size of the spectral data at different flow rates in different bands can be used to further improve the objective function of the multivariate linear regression in the existing technology.

[0089] In this embodiment, the objective function of the above optimization problem based on the regression model can be:

[0090] (5)

[0091] In formula (5), It indicates the minimum value of the flow rate used when measuring the substance of each anesthetic gas mixture sample, and it indicates the maximum value of the flow rate. Indicates the number of selected anesthetic gas mixture samples with different component contents. Indicates that in the training sample, The anesthetic gas mixture samples were tested at a gas flow rate of In the case of The component content of the anesthetic gas substance can be obtained by chemical analysis, for example, by gas chromatography. Indicates the Anesthetic gas substances to be solved dimensional coefficient vector, Indicates the The anesthetic gas mixture sample The composition content of the anesthetic gas substance is Time Maintenance positive coefficient matrix, Indicates the The anesthetic gas mixture samples were tested at a gas flow rate of Next, The characteristic band intervals of the anesthetic gas components dimensional spectral data absorbance vector, is the bias parameter to be solved, is the function value of the objective function. It should be noted that in the objective function expression, and The size relationship between them is such that the product of the three is ultimately a scalar, i.e. the final result It is also a scalar.

[0092] In some embodiments, The expression can be:

[0093] (6)

[0094] In formula (6), the The elements in the matrix are obtained by formula (4), because it is necessary to The composition of anesthetic gas substances is detected, so the number of bands used is indivual, Indicates that when the anesthetic gas mixture flow rate reaches hour, The first band in the band The contribution of the content of anesthetic gas substances, Indicates that when the anesthetic gas mixture flow rate reaches hour, In the band The first band (the last one) The contribution of each anesthetic gas substance to the content.

[0095] In this embodiment, different weight coefficients are assigned to the spectral absorbances in different bands under different flow rates to meet the variation differences of the spectral data under different flow rates, thereby improving the fitting accuracy of the regression model.

[0096] Step S150 : solving the objective function of the optimization problem to obtain an anesthetic gas estimation model, wherein the anesthetic gas estimation model is used to perform intelligent identification of anesthetic gas.

[0097] In this embodiment, the above-mentioned solving operation can be performed using calculation methods such as gradient descent and Newton's method, or can be performed using calculation tools such as MATLAB or R. MATLAB is a mathematical computing software with a rich optimization toolbox that can be used to solve various mathematical optimization problems, including parameter estimation and optimization of regression models. R is a statistical computing and data analysis software that also has many optimization packages and tools available for use, which can be used to solve optimization problems of regression models.

[0098] In this embodiment, by solving the objective function of the above-mentioned optimization problem, that is, when the above-mentioned objective function obtains a minimum value, the final regression model under different flow velocities can be obtained.

[0099] In this embodiment, the above-mentioned anesthetic gas estimation model may be:

[0100] (7)

[0101] In formula (7), Indicates the Anesthetic gas substances at flow rate Estimated values ​​of the following ingredients: Indicates the Within the characteristic band range, the spectral absorption rate vector corresponding to multiple bands, is the bias parameter that has been solved, Indicates the The anesthetic gas substances have been solved dimensional coefficient vector, Indicates the The anesthetic gas mixture sample The composition and content of various anesthetic gas substances.

[0102] In this embodiment, the anesthetic gas estimation model is an improved regression model that can intelligently identify and detect the components of the anesthetic gas in the existing anesthesia machine. The improved regression model is more accurate in estimating the components of each anesthetic substance, which helps doctors to make real-time adjustments to the concentration of anesthetic substances to meet the anesthesia needs during surgery.

[0103] In summary, the present invention provides an intelligent method for identifying anesthetic gas components. The method considers the differences in spectral data at different flow rates, and based on the distribution of the absorbance of the spectral data at different flow rates within the characteristic band interval, utilizes the skewness coefficient of the distribution, and corrects the content contribution in different bands through gamma transformation, thereby improving the accuracy of the contribution of the absorbance of the spectral data at different flow rates to each anesthetic gas substance. The objective function of the multiple linear regression model in the prior art is then improved based on the improved content contribution, so that the obtained improved regression model has better estimation accuracy for the spectral data at different flow rates.

[0104] Example 2:

[0105] See also Figure 4 , which shows a block diagram of an intelligent anesthetic gas component identification system provided by one embodiment of the present invention, the intelligent anesthetic gas component identification system specifically includes:

[0106] The acquisition module is used to acquire spectral data and corresponding flow rate data of multiple anesthetic gas mixture samples; wherein different anesthetic gas mixture samples have different flow rates.

[0107] The first building module is used to build a content contribution model of a single anesthetic gas based on the spectral data.

[0108] A modification module is used to modify the content contribution model of the single anesthetic gas based on the flow rate data to obtain a revised content contribution model; wherein the revised content contribution model takes into account the influence of the flow rate data on the spectral data; wherein the influence is used to indicate that when the flow rate data is higher than a certain threshold, the peak shape of the spectral data becomes narrower, and when the flow rate data is lower than a certain threshold, the peak shape of the spectral data becomes wider.

[0109] The second construction module is used to construct an optimization problem based on a regression model based on the modified content contribution model; wherein the objective function of the optimization problem is the difference between the actual content of a certain anesthetic gas actually detected and the estimated content calculated by the regression model.

[0110] A solution module is used to solve the objective function of the optimization problem, thereby obtaining an anesthetic gas estimation model, wherein the anesthetic gas estimation model is used to perform intelligent identification of anesthetic gas.

[0111] It should be noted that the order in which the embodiments of the present invention are described above is for illustrative purposes only and does not necessarily represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or sequential order shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0112] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on the differences from other embodiments.

Claims

1. An intelligent method for identifying anesthetic gas components, characterized in that: The method comprises: Acquiring spectral data and corresponding flow rate data of a plurality of anesthetic gas mixture samples, wherein different anesthetic gas mixture samples have different flow rates; Based on the spectral data, constructing a content contribution model of a single anesthetic gas; Based on the flow rate data, the content contribution model of the single anesthetic gas is modified to obtain a revised content contribution model; wherein the revised content contribution model takes into account the influence of the flow rate data on the spectral data; wherein the influence is used to indicate that when the flow rate data is higher than a certain threshold, the peak shape of the spectral data becomes narrower, and when the flow rate data is lower than a certain threshold, the peak shape of the spectral data becomes wider; Based on the modified content contribution model, an optimization problem based on a regression model is constructed; wherein the objective function of the optimization problem is the difference between the actual content of a certain anesthetic gas detected and the estimated content calculated by the regression model; Solving the objective function of the optimization problem to obtain an anesthetic gas estimation model, wherein the anesthetic gas estimation model is used to perform intelligent identification of anesthetic gas; The step of constructing a content contribution model of a single anesthetic gas based on the spectral data includes: Determining characteristic wavelength band intervals of each anesthetic gas in the anesthetic mixed gas sample based on the spectral data; Constructing a content contribution model of different bands within the characteristic band interval to a single anesthetic gas; wherein the content contribution model takes into account the response estimation value of the single anesthetic gas in different bands within the characteristic band interval; The expression of the content contribution model is: the content contribution model is expressed as the product of the proportional coefficient of the estimated absorption rate of any anesthetic gas in a certain band to the estimated absorption rates of the other anesthetic gases and the spectral absorption rate of the corresponding anesthetic gas in the band, wherein the product is subjected to maximum and minimum normalization processing.

2. The intelligent identification method of anesthetic gas components according to claim 1, characterized in that: The step of modifying the content contribution model of the single anesthetic gas based on the flow rate data to obtain a revised content contribution model includes: Determining an adaptive gamma value of the spectral data of the anesthetic gas mixture sample based on the flow rate data; wherein the adaptive gamma value is used to represent the difference between the skewness coefficient of the absorbance at the maximum flow rate and the skewness coefficient of the absorbance in each band within each characteristic band interval when the flow rate data is carried, wherein the difference is mapped by an exponential function; Based on the adaptive Gamma value, the content contribution model is corrected to obtain a corrected content contribution model.

3. The intelligent identification method of anesthetic gas components according to claim 1, characterized in that: The characteristic band interval is determined by the following method: For the spectral data of each anesthetic gas, determining the wavelength band corresponding to the maximum absorbance of the spectral data; Starting from the wavelength corresponding to the maximum absorbance of the spectral data, a sliding window of a preset length is slid to both sides by a distance of a number of sliding windows; The wavelength band within the sliding window and the wavelength band corresponding to the maximum value of the spectral data absorption rate are determined as the characteristic wavelength band interval.

4. The intelligent identification method of anesthetic gas components according to claim 1, characterized in that: In the expression of the objective function of the optimization problem, the actual content of the anesthetic gas actually detected is determined by the following method: The actual content of a certain anesthetic gas actually detected is represented by a preset training sample; wherein the training sample includes the component content of at least one anesthetic gas substance of at least one anesthetic mixed gas sample under at least one gas flow rate.

5. The intelligent identification method of anesthetic gas components according to claim 4, characterized in that: In the objective function of the optimization problem, the estimated content calculated by the regression model is determined by the following method: Construct the coefficient vector to be solved; The coefficient vector to be solved is multiplied by the correction coefficient matrix and the spectral data absorbance vector to obtain the product value; wherein the correction coefficient matrix is ​​the component content of any anesthetic gas substance in any anesthetic gas mixture sample, when the anesthetic gas flow rate data is a certain value ;Maintenance positive coefficient matrix; where, is the number of bands in the characteristic band interval, and the spectral data absorption rate vector is the spectral data absorption rate vector of the corresponding anesthetic gas substance in the characteristic band interval; The multiplied value is added to the bias parameter to be solved, and then at least the number of anesthetic mixed gas samples and the anesthetic gas flow rate data are summed to obtain the estimated content calculated by the regression model.

6. The intelligent identification method of anesthetic gas components according to claim 5, characterized in that: The objective function of the optimization problem is solved by using Lagrange number multiplication or gradient descent method.

7. The intelligent identification method of anesthetic gas components according to claim 5, characterized in that: The expression of the anesthetic gas estimation model is: The product of the solved coefficient vector, component content, and spectral absorptivity vector is added with the solved bias parameter; wherein the component content is the component content of the target anesthetic gas substance in the target anesthetic gas mixture sample; wherein the spectral absorptivity vector represents the spectral absorptivity vector corresponding to multiple bands within the characteristic band interval of the target anesthetic gas substance in the target anesthetic gas mixture sample.

8. An intelligent identification system for anesthetic gas components, characterized in that: The system comprises: An acquisition module, which is used to acquire spectral data and corresponding flow rate data of multiple anesthetic gas mixture samples; wherein different anesthetic gas mixture samples have different flow rates; A first building module is configured to build a content contribution model of a single anesthetic gas based on the spectral data; a modification module for modifying the content contribution model of the single anesthetic gas based on the flow rate data to obtain a modified content contribution model; wherein the modified content contribution model takes into account the influence of the flow rate data on the spectral data; wherein the influence is used to indicate that when the flow rate data is higher than a certain threshold, the peak shape of the spectral data becomes narrower, and when the flow rate data is lower than a certain threshold, the peak shape of the spectral data becomes wider; A second construction module is configured to construct an optimization problem based on a regression model based on the modified content contribution model; wherein the objective function of the optimization problem is the difference between the actual content of a certain anesthetic gas detected and the estimated content calculated by the regression model; a solving module, configured to solve the objective function of the optimization problem, thereby obtaining an anesthetic gas estimation model, wherein the anesthetic gas estimation model is configured to perform intelligent identification of the anesthetic gas; The first building block is specifically used to: Determining characteristic wavelength band intervals of each anesthetic gas in the anesthetic mixed gas sample based on the spectral data; Constructing a content contribution model of different bands within the characteristic band interval to a single anesthetic gas; wherein the content contribution model takes into account the response estimation value of the single anesthetic gas in different bands within the characteristic band interval; The expression of the content contribution model is: the content contribution model is expressed as the product of the proportional coefficient of the estimated absorption rate of any anesthetic gas in a certain band to the estimated absorption rates of the other anesthetic gases and the spectral absorption rate of the corresponding anesthetic gas in the band, wherein the product is subjected to maximum and minimum normalization processing.

Images