Online real-time quantitative detection method for iCa in laser-induced breakdown spectroscopy dialysate

Abstract

The invention relates to the technical field of solution electrolyte detection, in particular to an on-line real-time quantitative detection method for iCa in laser-induced breakdown spectroscopy dialyzate, which comprises the following steps: obtaining an original spectrum according to a training sample set, preprocessing the original spectrum to obtain a preprocessed spectrum, converting the preprocessed spectrum into a two-dimensional spectrum matrix, and calculating the iCa in the two-dimensional spectrum matrix; carrying out feature extraction to obtain a feature spectrum; the initial model is trained according to the characteristic spectrum and the first training sample set to obtain an optimization model; and establishing a regression curve by the optimization model according to the test sample set, and evaluating the optimization model to obtain a concentration prediction model. According to the invention, the original spectrum is pretreated to obtain the pretreated spectrum, so that the problems of non-uniform distribution of substances in the dialysate and non-uniform excitation degrees of the same sample in an experiment are solved; a concentration prediction model is obtained according to the regression curve, accurate detection of the content of the two elements of ionized calcium or potassium in the real flowing dialysate is achieved according to the concentration prediction model, and clinical requirements are completely met.

Description

Technical Field

[0001] The present invention relates to the technical field of solution electrolyte detection, and in particular to an online real-time quantitative detection method, device and storage medium of iCa in a dialysis fluid by laser induced breakdown spectroscopy. Background Art

[0002] Local citrate anticoagulation RCA has the advantages of fewer bleeding complications and longer filter life, and has become the preferred anticoagulation method for critically ill CRRT patients. The implementation principle of RCA is to chelate calcium ions with citrate and block the coagulation cascade to achieve the purpose of anticoagulation, so monitoring the level of ionized calcium (iCa) is the basis for ensuring the successful implementation of RCA.

[0003] During RCA, iCa in blood or dialysate needs to be frequently monitored, and bedside biochemical analyzers or traditional laboratory biochemical instruments are generally used. However, the extremely low in vitro iCa level (about 0.2-0.4 mmol / L) during RCA treatment not only exceeds the measurement range of traditional laboratory biochemical instruments, but also consumes a lot of manpower for frequent sampling, and the transportation of specimens and detection time are long; although bedside biochemical analyzers are convenient for extremely low iCa detection, they are only applicable to blood, and there are deviations in the determination of iCa in dialysate, and the cost of supporting consumables is huge, and at least several minutes of detection time is required; although access-type ion-selective electrodes can achieve the purpose of real-time detection, whether the components of ion-selective electrodes are toxic and the service life of the electrodes need to be observed for a long time, and they need to be recalibrated each time they are used. Therefore, there is currently no real-time online detection of iCa, which limits the clinical application of RCA-CRRT.

[0004] At present, the LIBS method is used to determine the content of trace metal elements in dialysate. The sample must be dried first and then made into a solid sample for determination. At the same time, the calcium concentration in the dialysate is low, the spectral signal is difficult to detect, and the total calcium and iCa are difficult to distinguish. Optimization is required to achieve the detection accuracy and precision that meet clinical needs. In terms of data processing, conventional univariate spectral data processing methods cannot meet the iCa detection requirements. Summary of the invention

[0005] The purpose of the present invention is to solve the shortcomings of the prior art and provide a method for online real-time quantitative detection of iCa in dialysate by laser induced breakdown spectroscopy, comprising the following steps: Obtaining an original spectrum according to a training sample set, preprocessing the original spectrum to obtain a preprocessed spectrum, converting the preprocessed spectrum into a two-dimensional spectrum matrix, and performing feature extraction according to the two-dimensional spectrum matrix to obtain a feature spectrum; The initial model is trained according to the characteristic spectrum and the first training sample set to obtain an optimized model; The optimization model establishes a regression curve according to the predicted values ​​of ionized calcium of the test sample set, and the optimization model is evaluated according to the regression curve to obtain a concentration prediction model.

[0006] Preferably, the step of obtaining an original spectrum according to a training sample set further comprises: Collecting and configuring the training sample set, and making a sample information table according to the training sample set; The laser pulse induces the training sample set to generate plasma, and the original spectrum is obtained according to the light emitted by the plasma.

[0007] Preferably, the laser pulse induces the training sample set to generate plasma, and the original spectrum is obtained according to the plasma emission light, further comprising: The laser pulse is incident on the training sample set to generate a stable plasma; The optical fiber collects the concentrated plasma emission light and couples the plasma emission light to the entrance of the spectrometer; A spectrometer disperses the plasma emitted light to form a spectrum; The detector collects the spectrum to obtain the original spectrum.

[0008] Preferably, preprocessing the original spectrum to obtain a preprocessed spectrum further comprises: Removing the baseline of the original image by wavelet transform filtering or morphological filtering to obtain a baseline-removed spectrum; The baseline-removed spectrum is normalized using the total area of ​​the continuous spectrum in the 430-530 nm band to obtain a normalized spectrum; The normalized spectrum is obtained by a sliding average method to obtain the preprocessed spectrum.

[0009] Preferably, extracting features from the preprocessed spectrum to obtain a feature spectrum further comprises: The preprocessed spectrum is converted into a two-dimensional spectrum matrix to obtain a two-dimensional training set, and an ion calcium iCa element concentration content label and a potassium ion element concentration content label corresponding to the two-dimensional training set spectrum matrix are formed according to the sample information table, wherein the row vector of the two-dimensional spectrum matrix corresponds to the spectral line intensity corresponding to the pixel value of the preprocessed spectrum, and the number of row vectors corresponds to the number of spectra in the two-dimensional spectrum matrix; Calculating the concentration correlation between the intensity of each spectral channel and the ionized calcium iCa concentration of the corresponding sample according to the two-dimensional training set, and sorting the spectral channels from high to low according to the concentration correlation to obtain a sorted training set; According to the concentration correlation, several spectral channels with high correlation are selected in the sorted training set as spectral features, and according to the spectral features, the pre-processed spectrum is reduced in dimension to obtain a reduced-dimension spectrum; The dimension-reduced spectrum is standardized to obtain the characteristic spectrum.

[0010] Preferably, the initial model is trained according to the characteristic spectrum to obtain an optimized model, further comprising: The initial model is trained according to the characteristic spectrum, the ion calcium iCa element concentration content label and the potassium ion element concentration content label of the first training sample set to obtain the optimization model, and the optimization model parameters are saved.

[0011] Preferably, the optimization model establishes a regression curve according to the predicted values ​​of ionized calcium of the test sample set, and evaluates the optimization model according to the regression curve to obtain a concentration prediction model, further comprising: The second characteristic spectrum of the test sample set tests the optimization model to obtain a test ion calcium iCa predicted value and a test potassium ion predicted value, a regression curve is established according to the test ion calcium iCa predicted value and the test potassium ion predicted value, and the optimization model is evaluated according to the regression curve to obtain a concentration prediction model.

[0012] Preferably, evaluating the optimization model according to the regression curve to obtain a concentration prediction model further comprises: If the mean square error or determination coefficient of the regression curve meets the set value, the optimization model parameters are saved to obtain a concentration prediction model; If the mean square error or determination coefficient of the regression curve does not meet the set value, the optimization model parameters are optimized and the prediction is performed again until the set value is met to obtain a concentration prediction model.

[0013] Based on the same concept, the present invention also provides a computer device, including a memory and a processor, wherein the memory stores computer-readable instructions, and when the computer-readable instructions are executed by the processor, the processor executes the steps of the method for online real-time quantitative detection of iCa in dialysate by laser induced breakdown spectroscopy as described in any of the embodiments.

[0014] Based on the same concept, the present invention also provides a storage medium storing computer-readable instructions. When the computer-readable instructions are executed by one or more processors, the one or more processors execute the steps of the online real-time quantitative detection method of iCa in dialysate by laser induced breakdown spectroscopy as described in any of the embodiments.

[0015] Compared with the prior art, the present invention has the following beneficial effects: The present invention obtains an original spectrum according to a training sample set, and pre-processes the original spectrum to obtain a pre-processed spectrum, thereby solving the problems of uneven distribution of substances in the dialysate and different excitation levels of the same sample in the experiment.

[0016] The present invention establishes a regression curve according to the predicted value of ionized calcium of the test sample set through the optimization model, and evaluates the optimization model according to the regression curve to obtain a concentration prediction model. According to the concentration prediction model, accurate detection of the contents of ionized calcium and potassium in real flowing dialysate is achieved, and clinical needs are fully met. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Various other advantages and benefits will become apparent to those of ordinary skill in the art by reading the following detailed description of the preferred embodiment.The drawings are only for the purpose of illustrating the preferred embodiments and are not to be construed as limiting the invention.

[0018] Figure 1 This is a flow chart of a method for online real-time quantitative detection of iCa in dialysate by laser induced breakdown spectroscopy of the present invention; Figure 2 Another flow chart of the on-line real-time quantitative detection method of iCa in dialysate by laser induced breakdown spectroscopy of the present invention; Figure 3 It is a sample pool diagram of the present invention; Figure 4 It is a diagram of the experimental device of the present invention; Figure 5 Spectra of the present invention, wherein (a) is the original spectrum, and (b) is the pre-treated spectrum of ionized calcium iCa; Figure 6 The model performance diagram of the present invention, wherein Figure (a) is the ion calcium iCa model performance diagram, and Figure (b) is the potassium ion model performance diagram.

[0019] 1. Laser; 2. First total reflection mirror; 3. Second total reflection mirror; 4. Diverging lens; 5. First converging lens; 6. Dichroic mirror; 7. Double glued lens; 8. Crescent lens; 9. Sample pool; 10. Power pump; 11. Dialysate pool; 12. Second converging lens; 13. Optical fiber; 14. Spectrometer; 91. Liquid inlet; 92. Liquid outlet; 93. Input end cross section; 94. Output end cross section. DETAILED DESCRIPTION

[0020] In order to make the purpose, technical scheme and advantages of the present invention clearer, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. Obviously, the described embodiments are part of the embodiments of the present application, rather than all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of this application.

[0021] Those skilled in the art will appreciate that, unless otherwise stated, the singular forms "a", "an", and "the" used herein may also include plural forms. It should be further understood that the term "comprising" used in the specification of the present invention refers to the presence of the features, integers, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or groups thereof.

[0022] Example 1 See also Figure 1 and Figure 2 As shown, this embodiment provides a method for online real-time quantitative detection of iCa in dialysate by laser induced breakdown spectroscopy, comprising the following steps: An original spectrum is obtained according to a training sample set, the original spectrum is preprocessed to obtain a preprocessed spectrum, the preprocessed spectrum is converted into a two-dimensional spectrum matrix, and feature extraction is performed according to the two-dimensional spectrum matrix to obtain a feature spectrum.

[0023] Preferably, obtaining the original spectrum according to the training sample set further comprises: A training sample set is collected and configured, and a sample information table is prepared according to the training sample set. Specifically, in this embodiment, a dialysate sample with an iCa gradient is collected and prepared. Considering that in real situations, in addition to iCa changes during CRRT, the potassium ion level in the dialysate also fluctuates significantly, a potassium ion concentration gradient is set in the sample at the same time. In order to conform to the actual sample situation and avoid introducing interference factors, the potassium ion concentration needs to be randomly increased during preparation, and it is tested to observe whether there is any mutual influence between ions.

[0024] Under laboratory conditions, according to the range of ionized calcium iCa of clinical dialysate during RCA-CRRT, real dialysate samples were collected as base solution, and calcium chloride was used as an adulterant. When collecting real dialysate samples, samples with low calcium / potassium content were required to be collected, and experimental samples were prepared with these samples as raw materials. In order to avoid the difference in the dialysate sample matrix caused by the addition of distilled water for dilution, and to reduce the influence of total calcium changes on the detection of ionized calcium iCa, dialysate with ionized calcium iCa levels in the normal range was selected as base solution, and 40% trisodium citrate solution and 2 mol / L potassium chloride solution were added to simulate the real situation of RCA-CRRT, and experimental samples with different ionized calcium iCa and potassium ion concentration gradients were prepared. That is, two substances are randomly doped into the base liquid for preparation, and the concentration gradient of the ionic calcium iCa element is designed to be 0.29-1.28 mmol / L, and the potassium is 3.8-8.3 mmol / L, with a total of 54 samples, of which 44 samples are used as training sets and 10 samples are used as test sets, and the test set has a similar data distribution to the training set.

[0025] In order to reduce the sedimentation of samples due to storage, make them more uniform, and better meet the actual operational needs of clinical dialysis fluid external circulation detection, the samples were prepared at the experimental site.

[0026] The configured samples were calibrated using the I-STAT blood gas analyzer to obtain the actual concentration of ionized calcium iCa of each sample.

[0027] A sample information table is prepared, the content of which includes the measured concentration of ionized calcium iCa of each sample. Specifically, in this embodiment, the sample information table needs to include the concentrations of two elements, ionized calcium iCa / potassium ions, of the training sample set.

[0028] The laser pulse induces the training sample set to generate plasma, and the original spectrum is obtained according to the plasma emission light. Specifically, in this embodiment, the LIBS experimental device is built, the flow module is designed, and the experimental parameters are optimized to collect the LIBS spectrum of the dialysate sample. For a single dialysate sample, a sufficiently large amount of LIBS spectrum data is collected, and a sufficiently long sample flow time is collected to obtain the original spectrum data set of the training sample.

[0029] The LIBS experimental device is set up, including the flow module, LIBS laser module and acquisition module.

[0030] See also Figure 3 As shown, the LIBS experimental device in this embodiment is constructed, including: a flow module, the flow module includes an ellipsoidal sample pool 9 for carrying a dialysate sample, and a power pump 10 for driving the dialysate sample in the ellipsoidal sample pool 9 to flow into a dialysate pool 11; LIBS laser module, the LIBS laser module includes a first total reflection mirror 2 for completely reflecting laser light, a second total reflection mirror 3, a laser diffusion lens group and a laser focusing lens group, the second total reflection mirror 3 is used to receive reflected laser light and generate high-intensity reflected laser light, the laser diffusion lens expands the high-intensity reflected laser light, the expanded high-intensity reflected laser light is focused on the dialysate in the ellipsoidal sample pool 9 through the laser focusing lens group, and the focused high-intensity reflected laser light generates plasma, specifically, in this embodiment, the laser 1 emits laser light, specifically, in this embodiment, the laser emission is controlled by the first total reflection mirror 2 and the second total reflection mirror 3, so as to prevent the laser pulse energy fluctuation caused by restarting when the measurement is interrupted for a short time; The acquisition module includes a dichroic mirror 6 disposed between the laser diffusion lens group and the laser focusing lens group. The reversely propagating plasma emission light is reflected by the dichroic mirror 6 and converged by the second converging lens. The converged plasma emission light is collected by the optical fiber 13 and coupled to the spectrometer 14. The LIBS spectrum of the plasma emission light dispersed by the spectrometer 14 is collected by the detector.

[0031] See also Figure 3and Figure 4 As shown, the ellipsoidal sample pool 9 is placed vertically, and the input end section 93 and the output end section 94 of the ellipsoidal sample pool 9 are located on the long semi-axis, and the input end 91 of the ellipsoidal sample pool 9 is higher than the output end 92 of the ellipsoidal sample pool. Specifically, in this embodiment, by placing the ellipsoidal sample pool vertically and arranging the input end and the output end of the ellipsoidal sample pool to be located on the long semi-axis of the ellipsoidal sample pool, the dialysate of the sample pool flows into the dialysate pool, and the wall of the sample pool 9 needs to be smooth and flat enough to ensure that the pool is clean and free of residues from the previous sample when the sample is replaced after each experiment.

[0032] Preferably, the minor axis dimension of the ellipsoidal sample pool 9 is 30 mm, the dimensions of the input end cross section 93 and the output end cross section 94 of the ellipsoidal sample pool are 15 mm, and the input end 91 and the output end 92 of the ellipsoidal sample pool 9 are smaller than the input end cross section 93 and the output end cross section 94. Specifically, in the present embodiment, the volume of the ellipsoidal sample pool 9 is smaller than that of other shapes in order to accurately monitor the electrolyte concentration in the dialysate in real time to avoid the generation of excessive lag effects and reduce the accuracy of real-time data feedback.

[0033] Preferably, the input end 91 of the ellipsoidal sample pool 9 is connected to the output end of the dialysate pool 11 through a medical tube, and the output end 92 of the ellipsoidal sample pool 9 is connected to the input end of the dialysate pool 11 through a medical tube, to ensure that the liquid in the pool flows fully without the occurrence of turbulence. Compared with cubes and cylinders, the ellipsoidal pool has a smaller surface area, is evenly stressed, and is more solid. In the experiment, the sample dialysate enters the sample pool 9 from the liquid inlet 91 on one side and flows out from the liquid outlet 92 on the other side, and the pool remains basically full of dialysate.

[0034] Preferably, the sample pool 9 is connected to the power pump 10 and the dialysate pool 11 through a medical pipeline, and the power pump 10 provides power to meet the flow rate of the sample in accordance with clinical dialysis, that is, 4 L / h; the total volume of the sample pool 9, the dialysate pool 11 and the dialysate 11 in all pipelines is 210 mL. The experimental measurement is carried out in the order of the sample ion calcium iCa concentration gradient from high to low, and distilled water is not used for cleaning when the sample is replaced. In actual operation, the previous sample is flowed into the dialysate pool by the power of the power pump 10. After emptying, the next sample is added and allowed to circulate in the entire loop for 10 minutes before the experiment is carried out. It is estimated that the relative effects of the previous sample on the ion calcium iCa and potassium ion K of the next sample are approximately 0.1% and 0.02%, respectively, which can be ignored; if distilled water is used to clean the liquid pipeline when the sample is replaced, the above effects will increase.

[0035] The designed container and connecting pipeline must ensure that they can be adequately cleaned when the sample is replaced after each experiment to avoid interference between samples of different concentrations. The power pump 10 should meet the flow rate of the sample in accordance with the clinical human flow rate. The optimization of experimental parameters should ensure that the spectral lines of ion calcium iCa 422.6nm and potassium ion 766.5nm are detected at the same time; at the same time, the laser energy is optimized to find the optimal solution of laser energy between reducing the energy to weaken the self-absorption of the potassium spectral line and increasing the energy to ensure spectral excitation. Optimization of experimental parameters also includes but is not limited to determining the optimization of the laser transverse excitation of the LIBS equipment, the transverse collection of the spectrometer, and the optimization of the spectrometer parameter settings.

[0036] The LIBS laser module further includes a laser 1, a first total reflection mirror 2, a second total reflection mirror 3, a divergent lens 4, a first converging lens 5, a double glued lens 7 and a crescent lens 8. In this example, the laser 1 generates a laser, and uses the first total reflection mirror 2 to reflect to the second total reflection mirror 3 to control the laser emission, so as to avoid stopping the laser operation when the measurement is interrupted for a short time, and causing the laser pulse energy to fluctuate when it is restarted. If the laser 1 has the function of generating a stable laser pulse immediately after being turned on, it is also possible not to reflect to the second total reflection mirror 3 through the first total reflection mirror 2 to control the laser emission. Reflection from the first total reflection mirror 2 to the second total reflection mirror 3 can play a good role in controlling the laser emission. The use of an ellipsoidal pool has a small surface area compared to a cube and a cylinder, and the force is evenly applied, which is more solid.

[0037] The divergent lens 4, the first converging lens 5, the double-cemented lens 7 and the crescent lens 8 realize the effect of converging the laser pulse at a large angle in the liquid and focusing it on a light spot as small as possible, which can be called a laser large-angle converging and focusing module. In this example, the beam cross-sectional diameter is first expanded by 4 times from 6mm to 24mm through a combination of a divergent lens 4 (such as f=-50mm) and a first converging lens 5 (such as f=200mm), and then the laser pulse is focused on the dialysate in the sample pool through the double-cemented lens 7 and the crescent lens 8; the combination of the double-cemented lens 7 and the crescent lens 8 reduces spherical aberration, shortens the focal length to 37.5mm, increases the beam convergence angle, and ensures the excitation of a stable plasma in the liquid. In addition, in this example, the laser operating wavelength is 1064nm, the repetition frequency is 5Hz, and the laser pulse width is 16ns.

[0038] The acquisition module includes a dichroic mirror 6, a second converging lens 12, an optical fiber 13 and a spectrometer 14. After the laser pulse is focused, a plasma is generated in the liquid. The plasma emission light propagates in the opposite direction along the incident axis of the laser pulse, and is reflected by the dichroic mirror 6 and converged at its focus by a focusing lens 12 (f = 50mm). An optical fiber 13 is placed there to collect the converged plasma emission light and couple it to the entrance of a spectrometer 14. The collected plasma emission light is dispersed by the spectrometer 14, and the detector collects the spectrum. In this example, a medium-bandwidth spectrometer (Mechelle 5000, Andor Technology) is selected, with a spectral range of 220-900nm and a resolution of λ / (Δλ) = 5000, which meets the requirements for synchronous detection of the electrolyte elements of interest. The spectrometer is coupled to an enhanced CCD camera (ICCD, iStar, Andor Technology) to improve the signal-to-noise ratio. This choice is only an example and is not limited to this. For example, a compact multi-channel spectrometer coupled to a CCD camera can also be used.

[0039] A sufficiently large amount of LIBS spectral data should be collected as much as possible while the sample is flowing to avoid the impact of sample inhomogeneity and obtain the overall information of the sample.

[0040] When collecting spectra in the experiment, the optimized experimental parameters were: laser energy of 71.5 mJ, spectrometer parameters were set to delay of 150 ns, gate width of 3000 ns, gain of 4095, the spectrum was the accumulation of 20 plasma excitation signals, and 102 spectra were collected for each sample.

[0041] In order to avoid the disturbance of the wavefront of the incident laser pulse and the outgoing plasma emission light caused by the liquid surface fluctuation when the laser pulse is incident through the free surface between the liquid and the air, the laser is incident through the contact surface of the liquid and the transparent wall of the sample cell, and the plasma emission light is detected in the opposite direction on the same axis.

[0042] In order to avoid the disturbance of the wave surface of the incident laser and the outgoing plasma emission light caused by the bubbles left over from laser ablation in the liquid, the laser pulse incidence and the plasma emission light detection are carried out in the horizontal axis.

[0043] After the laser pulse is focused, plasma is generated inside the liquid. The plasma emission light propagates in the opposite direction of the incident axis of the laser pulse, and is focused near its focus by a focusing lens (f=50 mm) after being reflected by a dichroic mirror. An optical fiber is placed here to collect the converged plasma emission light and couple it to the entrance of a spectrometer. The collected plasma emission light is dispersed by the spectrometer, and the detector collects the spectrum.

[0044] In this work, a broadband spectrometer (Mechelle 5000, Andor Technology) was selected with a spectral range of 220-900 nm and a resolution of λ / (Δλ)=5000 to meet the requirements for simultaneous detection of the electrolyte elements of interest. The spectrometer was coupled to an enhanced CCD camera (ICCD, iStar, Andor Technology) to improve the signal-to-noise ratio of the spectral lines.

[0045] Preferably, the laser pulse induces the training sample set to generate plasma, and the original spectrum is obtained according to the plasma emission light, further comprising: Laser pulses were irradiated into the training sample set to produce a stable plasma; The optical fiber collects the concentrated plasma emission light and couples the plasma emission light to the entrance of the spectrometer; The spectrometer disperses the plasma emission light to form a spectrum; The detector collects the spectrum to obtain the original spectrum.

[0046] Preferably, preprocessing the original spectrum to obtain a preprocessed spectrum further comprises: Removing the baseline of the original image by wavelet transform filtering or morphological filtering to obtain a baseline-removed spectrum; The normalized spectrum is obtained by normalizing the spectrum without the baseline using the total area of ​​the continuous spectrum in the 430-530 nm band. Specifically, in this embodiment, refer to Figure 5 As shown in (a), for each original spectrum in the spectral data set of the first training sample set, continuous spectrum normalization is performed. Continuous spectrum normalization is an intensity normalization process using the continuous spectrum in the 430-530 nm band. The reason for adopting this method is that in a spectrum, the main element emission lines come from calcium and potassium, and the concentrations of these two elements are constantly changing. In comparison, the continuous spectrum in the 430-530 nm band is a relatively constant quantity and can be used as a normalization standard. The specific operation process is that for each spectrum, the spectral intensity values ​​corresponding to all pixels (wavelengths) in the 430-530 nm spectrum are summed to obtain the intensity sum, and the spectral intensity values ​​corresponding to all pixels (wavelengths) of this spectrum are divided by this intensity sum to perform intensity normalization to obtain a normalized spectrum. The normalized spectrum is obtained by a sliding average method to obtain a pre-processed spectrum. Specifically, in this embodiment, refer to Figure 5 As shown in (b), for each LIBS spectrum collected from the dialysate sample, the sliding average method is used to obtain the pre-processed spectrum. The sliding average method uses an average processing with a step size of 5 and every 40 spectra are averaged as one to obtain the pre-processed spectrum.

[0047] Preferably, extracting features from the preprocessed spectrum to obtain a feature spectrum further comprises: The preprocessed spectrum is converted into a two-dimensional spectrum matrix to obtain a two-dimensional training set, and the ion calcium iCa element concentration content label and the potassium ion element concentration content label corresponding to the two-dimensional training set spectrum matrix are formed according to the sample information table, wherein the row vector of the two-dimensional spectrum matrix corresponds to the spectral line intensity corresponding to the pixel value of the preprocessed spectrum, and the number of row vectors corresponds to the number of spectra in the two-dimensional spectrum matrix, and the preprocessed spectrum is converted into a two-dimensional spectrum matrix for data calculation; The concentration correlation between the intensity of each spectral channel and the ion calcium iCa concentration of the corresponding sample is calculated according to the two-dimensional training set, and the spectral channels are sorted from high to low according to the concentration correlation to obtain a sorted training set. Specifically, in this embodiment, the two-dimensional spectral matrix of the two-dimensional training set is calculated by feature selection to obtain the correlation between each feature and the corresponding calcium / potassium element concentration, and the SelectKBest algorithm is used to calculate the one-dimensional spectral feature correlation vector. This one-dimensional correlation vector corresponds to the corresponding pixel (wavelength) vector to obtain feature information. This feature information is sorted according to the correlation level, and some features with the highest correlation are selected. The feature selection calculation adopts methods including Pearson coefficient calculation, mutual information calculation, f-regression calculation, etc. to calculate the correlation, so as to select the top 200 features; According to the concentration correlation, several spectral channels with high correlation are selected as spectral features in the sorted training set. According to the spectral features, the pre-processed spectrum is reduced in dimension to obtain a reduced-dimension spectrum. The number of features with the highest correlation is selected, and generally an integer between 100 and 300 can be selected. The dimension-reduced spectrum is normalized to obtain a characteristic spectrum. Specifically, in this embodiment, [0, 1] interval normalization may be used, and the normalization parameters are transferred to the training sample set spectrum.

[0048] The initial model is trained according to the characteristic spectrum and the first training sample set to obtain an optimized model. Specifically, in this embodiment, under the same clinical conditions and experimental conditions, real dialysate samples (calcium / potassium content range) are collected and configured for model testing.

[0049] Preferably, the initial model is trained according to the characteristic spectrum to obtain an optimized model, further comprising: The initial model is trained according to the characteristic spectrum, the ion calcium iCa element concentration content label and the potassium ion element concentration content label of the first training sample set to obtain an optimized model, and the optimized model parameters are saved.

[0050] The optimization model establishes a regression curve according to the predicted values ​​of ionized calcium of the test sample set, and the optimization model is evaluated according to the regression curve to obtain a concentration prediction model.

[0051] Preferably, the optimization model establishes a regression curve according to the predicted values ​​of ionized calcium of the test sample set, and evaluates the optimization model according to the regression curve to obtain a concentration prediction model, further comprising: The second characteristic spectrum of the test sample set is used to test the optimization model to obtain the test ion calcium iCa predicted value and the test potassium ion predicted value, a regression curve is established according to the test ion calcium iCa predicted value and the test potassium ion predicted value, and the optimization model is evaluated according to the regression curve to obtain a concentration prediction model.

[0052] Preferably, evaluating the optimization model according to the regression curve to obtain a concentration prediction model further comprises: If the mean square error or determination coefficient of the regression curve meets the set value, the optimized model parameters are saved to obtain the concentration prediction model; If the mean square error or determination coefficient of the regression curve does not meet the set value, the model parameters are optimized and predicted again until the set value is met to obtain the concentration prediction model.

[0053] Under the same experimental conditions and experimental parameters, collect the LIBS spectra of the dialysate samples of the test set, preprocess the LIBS spectrum data of the test set samples to obtain the preprocessed spectrum, extract the features of the test set spectrum matrix, obtain the characteristic spectrum of the test set, standardize the characteristic spectrum matrix of the test set to generate the characteristic spectrum, and input it into the optimization model. Please refer to Figure 6 As shown in Table 1, the predicted concentration values ​​of two elements, ionized calcium iCa / potassium, of the test set samples were obtained. When designing the number of samples, multiple test sets need to be designed to ensure that the stable prediction performance of the model can be verified during the test process, and the prediction performance index parameters of the model are obtained, as shown in Table 1.

[0054] Table 1: Performance parameters of the model obtained by the preferred method In real clinical applications, the experimental device is embedded in the dialysate circulation system to collect the patient's dialysate LIBS spectra, using the above-mentioned spectral preprocessing, feature extraction and dimensionality reduction, and spectral standardization, the obtained characteristic spectra are input into the concentration prediction model for prediction, and the patient's dialysate ion calcium iCa / potassium content is output respectively. The prediction performance index parameters of the above-mentioned model are used to characterize the accuracy performance of the determination of two elements in the dialysate of clinical patients.

[0055] Example 2 In some embodiments of the present application, a computer device is also provided, including a memory and a processor, wherein the memory stores computer-readable instructions, and when the computer-readable instructions are executed by the processor, the processor performs the steps of the online real-time quantitative detection method of iCa in dialysate by laser induced breakdown spectroscopy as described in any one of Example 1.

[0056] The present invention also provides a storage medium storing computer-readable instructions, which, when executed by one or more processors, enables the one or more processors to perform the steps of the laser-induced breakdown spectroscopy online real-time quantitative detection method of iCa in dialysate as described in any one of Example 1.

[0057] It is understandable that, for the aforementioned laser-induced breakdown spectroscopy iCa online real-time quantitative detection method in dialysate, if it is implemented in the form of a software function module and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention is essentially or part of the contribution to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer server, or a network device, etc.) to perform all or part of the steps of the various embodiments of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk and other media that can store program codes.

[0058] Computer readable storage media may include data signals propagated in baseband or as part of a carrier wave, wherein readable program codes are carried. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. The readable storage medium may also be any readable medium other than a readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, device, or device. The program codes contained on the readable storage medium may be transmitted using any suitable medium, including but not limited to wireless, wired, optical cable, RF, etc., or any suitable combination thereof.

[0059] The above is only a preferred embodiment of the present invention, and the protection scope of the present invention is not limited to the above embodiments. All technical solutions under the concept of the present invention belong to the protection scope of the present invention. It should be pointed out that for ordinary technicians in this technical field, some improvements and modifications without departing from the principle of the present invention should also be regarded as the protection scope of the present invention.

Claims

1. A method for online real-time quantitative detection of iCa in dialysate by laser induced breakdown spectroscopy, characterized in that: The following steps are involved: Obtaining an original spectrum according to a training sample set, preprocessing the original spectrum to obtain a preprocessed spectrum, converting the preprocessed spectrum into a two-dimensional spectrum matrix, and performing feature extraction according to the two-dimensional spectrum matrix to obtain a feature spectrum; The initial model is trained according to the characteristic spectrum and the first training sample set to obtain an optimized model; The optimization model establishes a regression curve according to the predicted values ​​of ionized calcium of the test sample set, and the optimization model is evaluated according to the regression curve to obtain a concentration prediction model.

2. The method for online real-time quantitative detection of iCa in dialysate by laser induced breakdown spectroscopy according to claim 1, characterized in that: The step of obtaining the original spectrum according to the training sample set further comprises: Collecting and configuring the training sample set, and making a sample information table according to the training sample set; The laser pulse induces the training sample set to generate plasma, and the original spectrum is obtained according to the light emitted by the plasma.

3. The method for online real-time quantitative detection of iCa in dialysate by laser induced breakdown spectroscopy according to claim 2, characterized in that: The laser pulse induces the training sample set to generate plasma, and the original spectrum is obtained according to the plasma emission light, further comprising: The laser pulse is incident on the training sample set to generate a stable plasma; The optical fiber collects the concentrated plasma emission light and couples the plasma emission light to the entrance of the spectrometer; A spectrometer disperses the plasma emitted light to form a spectrum; The detector collects the spectrum to obtain the original spectrum.

4. The method for online real-time quantitative detection of iCa in dialysate by laser induced breakdown spectroscopy according to claim 3, characterized in that: Preprocessing the original spectrum to obtain a preprocessed spectrum further includes: Removing the baseline of the original image by wavelet transform filtering or morphological filtering to obtain a baseline-removed spectrum; The baseline-removed spectrum is normalized using the total area of ​​the continuous spectrum in the 430-530 nm band to obtain a normalized spectrum; The normalized spectrum is obtained by a sliding average method to obtain the preprocessed spectrum.

5. The method for online real-time quantitative detection of iCa in dialysate by laser induced breakdown spectroscopy according to claim 4, characterized in that: Extracting features from the preprocessed spectrum to obtain a feature spectrum further comprises: The preprocessed spectrum is converted into a two-dimensional spectrum matrix to obtain a two-dimensional training set, and an ion calcium iCa element concentration content label and a potassium ion element concentration content label corresponding to the two-dimensional training set spectrum matrix are formed according to the sample information table, wherein the row vector of the two-dimensional spectrum matrix corresponds to the spectral line intensity corresponding to the pixel value of the preprocessed spectrum, and the number of row vectors corresponds to the number of spectra in the two-dimensional spectrum matrix; Calculating the concentration correlation between the intensity of each spectral channel and the ionized calcium iCa concentration of the corresponding sample according to the two-dimensional training set, and sorting the spectral channels from high to low according to the concentration correlation to obtain a sorted training set; According to the concentration correlation, several spectral channels with high correlation are selected in the sorted training set as spectral features, and according to the spectral features, the pre-processed spectrum is reduced in dimension to obtain a reduced-dimension spectrum; The dimension-reduced spectrum is standardized to obtain the characteristic spectrum.

6. The method for online real-time quantitative detection of iCa in dialysate by laser induced breakdown spectroscopy according to claim 5, characterized in that: The initial model is trained according to the characteristic spectrum to obtain an optimized model, further comprising: The initial model is trained according to the characteristic spectrum, the ion calcium iCa element concentration content label and the potassium ion element concentration content label of the first training sample set to obtain the optimization model, and the optimization model parameters are saved.

7. The method for online real-time quantitative detection of iCa in dialysate by laser induced breakdown spectroscopy according to claim 6, characterized in that: The optimization model establishes a regression curve according to the predicted value of ionized calcium of the test sample set, and evaluates the optimization model according to the regression curve to obtain a concentration prediction model, further comprising: The second characteristic spectrum of the test sample set tests the optimization model to obtain a test ion calcium iCa predicted value and a test potassium ion predicted value, a regression curve is established according to the test ion calcium iCa predicted value and the test potassium ion predicted value, and the optimization model is evaluated according to the regression curve to obtain a concentration prediction model.

8. The method for online real-time quantitative detection of iCa in dialysate by laser induced breakdown spectroscopy according to claim 7, characterized in that: Evaluating the optimization model according to the regression curve to obtain a concentration prediction model further comprises: If the mean square error or determination coefficient of the regression curve meets the set value, the optimization model parameters are saved to obtain a concentration prediction model; If the mean square error or determination coefficient of the regression curve does not meet the set value, the optimization model parameters are optimized and the prediction is performed again until the set value is met to obtain a concentration prediction model.

9. A computer device, characterized in that: It comprises a memory and a processor, wherein the memory stores computer-readable instructions, and when the computer-readable instructions are executed by the processor, the processor executes the steps of the method for online real-time quantitative detection of iCa in dialysate by laser induced breakdown spectroscopy as described in any one of claims 1 to 8.

10. A storage medium storing computer-readable instructions, characterized in that: When the computer-readable instructions are executed by one or more processors, the one or more processors execute the steps of the method for online real-time quantitative detection of iCa in dialysate by laser induced breakdown spectroscopy according to any one of claims 1 to 8.

Images