Quantitative-discriminative integrated uremia intelligent detection system

Abstract

The application discloses a quantitative-discriminative integrated uremia intelligent detection system, and relates to the field of medical spectrum analysis.The system comprises a spectrum collection module, which is used for collecting near-infrared transmission spectrum data of a serum sample to be measured; and a double-model detection module, which is used for determining urea concentration and a uremia state of the serum sample to be measured by using a pre-constructed urea concentration quantitative regression model and a uremia state discriminative classification model respectively according to the near-infrared transmission spectrum data.The application realizes accurate determination of urea concentration and rapid automatic identification of uremia, and has significant advantages in reducing cost, simplifying process and improving diagnosis efficiency.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of medical spectrum analysis, in particular to a quantitative-discriminative integrated uremia intelligent detection system. BACKGROUND

[0002] Uremia is a syndrome caused by the accumulation of metabolic waste in the body due to severe renal failure, and its clinical manifestations involve electrolyte disorders, acid-base imbalance, metabolic abnormalities and multiple organ dysfunction. Since the disease often lacks specific symptoms in the early stage, patients are usually diagnosed only when the disease has progressed to the late stage, so it is urgent to develop a rapid, sensitive and convenient detection technology to realize timely diagnosis and dynamic monitoring. Serum urea concentration is a key biochemical indicator for evaluating protein metabolism status and kidney function level. The current clinical detection methods include gas chromatography, ion conductivity and urease method, etc. Although the above methods have high sensitivity and accuracy, they all have problems such as complicated sample pretreatment steps, long determination period, dependence on chemical reagents and complex instrument operation, which limit their application in clinical high-throughput screening. Therefore, the existing technology still has certain limitations in the specific diagnosis of uremia, and it is difficult to meet the dual needs of rapid detection and disease recognition.

[0003] Near-infrared spectroscopy technology has shown a wide application prospect in body fluid analysis and disease monitoring due to its non-destructive, no need for chemical reagents, fast response, high detection efficiency and other advantages. It does not need to pretreat the serum sample, and the measurement process can be completed within one minute, which is particularly suitable for clinical rapid screening and dynamic monitoring. At the same time, near-infrared spectroscopy carries a large amount of chemical and structural information, and has unique advantages in the analysis of multi-component mixed systems. However, due to the influence of complex components and significant matrix effect of serum samples, the near-infrared modeling process still faces the problems of insufficient model generalization ability and unstable prediction accuracy, which is difficult to meet the requirements of detection performance in clinical practical application.

[0004] Therefore, in order to achieve higher detection accuracy and stronger model generalization ability, it is urgent to optimize and improve the spectrum pretreatment strategy, feature variable selection method and modeling algorithm, etc. SUMMARY

[0005] The purpose of the present application is to provide a quantitative-discriminative integrated uremia intelligent detection system, which can realize accurate determination of urea concentration and rapid automatic identification of uremia.

[0006] To achieve the above purpose, the present application provides the following solutions:

[0007] The present application provides a quantitative-discriminative integrated uremia intelligent detection system, which comprises:

[0008] a spectrum acquisition module, configured to acquire near-infrared transmission spectrum data of a serum sample to be tested;

[0009] a double-model detection module, configured to determine urea concentration and uremia state of the serum sample to be tested according to the near-infrared transmission spectrum data by using a pre-constructed urea concentration quantitative regression model and a uremia state discriminant classification model respectively.

[0010] In an embodiment, the spectrum acquisition module acquires the near-infrared transmission spectrum data of the serum sample to be tested under constant temperature condition; the serum sample to be tested is located in a cuvette, and the cuvette is placed in a constant-temperature sample holder.

[0011] In an embodiment, the double-model detection module comprises: a spectrum processing submodule, configured to generate model input data according to the near-infrared transmission spectrum data; the model input data comprises absorption intensity values of the near-infrared transmission spectrum data at each key characteristic wavelength; a quantitative detection submodule, configured to determine urea concentration of the serum sample to be tested by using the pre-constructed urea concentration quantitative regression model according to the model input data; and a state discrimination submodule, configured to determine uremia state of the serum sample to be tested by using the pre-constructed uremia state discriminant classification model according to the model input data.

[0012] In an embodiment, the spectrum processing submodule comprises: a spectrum transformation unit, configured to perform continuous wavelet transform on the near-infrared transmission spectrum data, so as to improve spectral resolution; and a wavelength screening unit, configured to perform wavelength screening on the near-infrared transmission spectrum data after continuous wavelet transform by using a competitive adaptive reweighted sampling method, so as to obtain key characteristic wavelengths.

[0013] an input data construction submodule, configured to determine model input data according to the key characteristic wavelengths and the near-infrared transmission spectrum data.

[0014] In an embodiment, the spectrum transformation unit performs continuous wavelet transform on the near-infrared transmission spectrum data by using a Symmlet filter with vanishing moments of 4.

[0015] In an embodiment, the urea concentration quantitative regression model is constructed by using a partial least squares method according to a first sample set in advance; the uremia state discriminant classification model is constructed by using a partial least squares method according to a second sample set in advance; the first sample set comprises near-infrared transmission spectrum data of multiple serum samples and urea concentration of each serum sample; and the second sample set comprises near-infrared transmission spectrum data of multiple serum samples and uremia state of each serum sample.

[0016] In an embodiment, the multiple serum samples comprise serum samples of normal people and serum samples of uremia patients.

[0017] In an embodiment, the uremia state is 1 or -1, 1 representing positive and -1 representing negative.

[0018] In an embodiment, the spectrum acquisition module and the double model detection module are both deployed in a computer device.

[0019] In an embodiment, the system further comprises a visualization module for displaying the urea concentration and the uremia state of the serum sample to be tested to assist a doctor in diagnosing uremia.

[0020] According to the specific embodiments provided in the present application, the present application has the following technical effects: the present application provides a quantitative-discriminative integrated uremia intelligent detection system, which only needs to collect near-infrared transmission spectrum data of a serum sample to be tested, can accurately detect urea concentration and identify uremia state without relying on chemical reagents and complex operations, realizes accurate determination of urea concentration and rapid automatic identification of uremia, and has significant advantages in reducing cost, simplifying process and improving diagnosis efficiency. BRIEF DESCRIPTION OF DRAWINGS

[0021] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed in the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0022] Figure 1 A functional module schematic diagram of a quantitative-discriminative integrated uremia intelligent detection system according to an embodiment of the present application.

[0023] Figure 2 A schematic diagram of the construction process of a double model in an embodiment of the present application.

[0024] Figure 3 A near-infrared transmission spectrum schematic diagram of a serum sample in an embodiment of the present application.

[0025] Figure 4 A high-resolution spectrum schematic diagram obtained after continuous wavelet transform processing in an embodiment of the present application.

[0026] Figure 5 A graph showing the change relationship of the root mean square error of the urea concentration quantitative model for the calibration set, the validation set and the prediction set with the number of partial least squares regression factors in an embodiment of the present application.

[0027] Figure 6 A correlation schematic diagram between the predicted value and the reference value of the urea concentration quantitative model in an embodiment of the present application.

[0028] Figure 7 A root mean square error of the uremia state discriminant classification model in an embodiment of the present application for the calibration set, the validation set, and the prediction set with respect to the change of the partial least squares regression factor number.

[0029] Figure 8 A prediction classification result distribution diagram of the uremia state discriminant classification model in an embodiment of the present application for the uremia sample and the healthy sample. DETAILED DESCRIPTION

[0030] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of protection of the present application.

[0031] The present application systematically integrates near-infrared spectroscopy technology and chemometrics modeling strategy, and constructs high-precision urea concentration quantitative regression models and uremia state discriminant classification models for the spectral feature information in serum samples that is significantly related to urea concentration and uremia state, so as to realize accurate determination of urea indicators in serum and intelligent discrimination of uremia, complete synchronous analysis of biochemical index extraction and disease recognition, and provide efficient, stable, and reliable technical support for intelligent auxiliary diagnosis of clinical kidney function impairment.

[0032] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the present application will be further described in detail below with reference to the drawings and specific embodiments.

[0033] In an exemplary embodiment, as shown in Figure 1 A quantitative-discriminant integrated uremia intelligent detection system is provided, which is deployed in a computer device and executed by the computer device. The system includes a spectrum acquisition module 101 and a double-model detection module 102.

[0034] The spectrum acquisition module 101 is configured to acquire near-infrared transmission spectrum data of a serum sample to be measured.

[0035] Specifically, the spectrum acquisition module 101 acquires near-infrared transmission spectrum data of a serum sample to be measured under constant temperature conditions. The serum sample to be measured is located in a cuvette, and the cuvette is placed in a constant-temperature sample holder.

[0036] The double-model detection module 102 is configured to determine the urea concentration and the uremia state of the serum sample to be detected according to the near-infrared transmission spectrum data by using a pre-constructed urea concentration quantitative regression model and a uremia state discriminant classification model, respectively.

[0037] Specifically, the urea concentration quantitative regression model is constructed by using a partial least squares method according to a first sample set.

[0038] The first sample set includes near-infrared transmission spectrum data of a plurality of serum samples and urea concentrations of the serum samples, and the second sample set includes near-infrared transmission spectrum data of a plurality of serum samples and uremia states of the serum samples.

[0039] The plurality of serum samples include serum samples of normal people and serum samples of uremia patients.

[0040] In one specific application example, the double-model detection module 102 includes a spectrum processing submodule, a quantitative detection submodule, and a state discrimination submodule.

[0041] The spectrum processing submodule is configured to generate model input data according to the near-infrared transmission spectrum data. The model input data includes absorption intensity values of the near-infrared transmission spectrum data at each key characteristic wavelength.

[0042] Specifically, the spectrum processing submodule includes a spectrum transformation unit, a wavelength screening unit, and an input data construction submodule.

[0043] The spectrum transformation unit is configured to perform continuous wavelet transform on the near-infrared transmission spectrum data. The spectrum transformation unit uses a Symmlet filter with a vanishing moment of 4 to perform continuous wavelet transform on the near-infrared transmission spectrum data.

[0044] The wavelength screening unit is configured to perform wavelength screening on the near-infrared transmission spectrum data after continuous wavelet transform by using a competitive adaptive reweighted sampling method to obtain key characteristic wavelengths.

[0045] The input data construction submodule is configured to determine model input data according to the key characteristic wavelengths and the near-infrared transmission spectrum data.

[0046] The quantitative detection submodule is configured to determine the urea concentration of the serum sample to be detected according to the model input data by using a pre-constructed urea concentration quantitative regression model.

[0047] The state discrimination sub-module is configured to determine the uremia state of the serum sample to be tested according to the model input data and by using a pre-constructed uremia state discrimination classification model.

[0048] In another exemplary embodiment, the quantitative-discriminative integrated uremia intelligent detection system further comprises a visualization module 103. The visualization module 103 is configured to display the urea concentration and the uremia state of the serum sample to be tested, so as to assist doctors in diagnosing uremia.

[0049] In the present application, as shown in Figure 2 , the construction process of the urea concentration quantitative regression model and the uremia state discrimination classification model is as follows.

[0050] 1) Collect serum samples of normal people and uremia patients, and collect important index information such as urea in the corresponding test sheets.

[0051] Among them, the number of normal serum samples and uremia patient serum samples is not less than 20 respectively, to ensure that the samples are representative, and the metabolic index test sheets are obtained synchronously. The urea concentration range of the selected serum samples is 0-50 mM. Combined with the test results and the suggestions of clinicians, it is determined that the urea concentration is the main diagnostic basis, and the normal concentration range is 2.6-7.5 mM, which can be preliminarily judged as abnormal if it exceeds this range.

[0052] 2) Measure the near-infrared transmission spectrum data of each serum sample, and randomly divide the near-infrared transmission spectrum data of the serum sample into a training set and a test set according to a ratio of 3:1.

[0053] Specifically, 0.2 mL of serum sample is placed in a quartz cuvette with an optical path of 1 mm and placed in a 30°C constant temperature sample holder. After the temperature is stabilized, the near-infrared transmission spectrum is collected. The number of spectral scanning is set to 64, and the spectral wavenumber point interval after Fourier transform is 2 cm -1 . The spectral acquisition time of a single sample is about 1 minute. The near-infrared transmission spectrum in the experiment is shown in Figure 3 , and in the obtained original near-infrared transmission spectrum, the serum sample has two typical water strong absorption peaks near 6900 cm -1 and 5200 cm -1 . The peak intensity of 5200 cm -1 has exceeded the linear detection range of the instrument, so this wave band is not used in subsequent modeling. Overall, the spectral baseline of the uremia sample is higher than that of the healthy sample, and due to the strong water absorption, the spectral peaks are severely overlapped, making it difficult to directly distinguish other biological molecule signals.

[0054] 3) Preprocess the near-infrared transmission spectrum data.

[0055] Specifically, the collected spectral wavelength range is divided into the water molecule and the biological molecule absorption band range according to the molecular absorption characteristics. Spectra of the full wavelength range, the water molecule absorption wavelength range and the biological molecule absorption wavelength range are selected respectively for modeling and analysis.

[0056] In order to identify the characteristic waveband closely related to urea and uremia state, the near-infrared transmission spectrum data of healthy people and the near-infrared transmission spectrum data of uremia patients are compared first, and the significant difference characteristic peak is observed.

[0057] The main absorption characteristics of the spectrum are as follows: the spectrum based on Fourier near-infrared spectrometer scanning, the wave number range is 12000~4000cm -1 , the absorption of water molecules is mainly concentrated in 7500~6500cm -1 , the frequency doubling absorption of NH group in urea and other primary amine molecules is about 6700cm -1 , the frequency doubling absorption of NH group in secondary amide in protein is about 6400cm -1 , the combination frequency of amide is about 4600 and 4850cm -1 , the first frequency doubling absorption of CH group is 5900~5700cm -1 , and the combination frequency absorption of CH group is 4500~4200cm -1 . In summary, the characteristic absorption waveband of water molecules is 7500~6500cm -1 , the characteristic absorption waveband of urea and other metabolism related molecules is concentrated in 6500~5700cm -1 and 4900~4200cm -1 , and thus can be used as the key characteristic waveband for rapid discrimination of uremia.

[0058] In order to separate the overlapping peaks, enhance the characteristic signal of biological molecules, and reduce the interference of baseline drift on quantitative analysis, the continuous wavelet transform method is used to process the near-infrared transmission spectrum data, enhance the resolution of the spectrum, and thus improve the signal-to-noise ratio of the spectrum, improve the detection sensitivity and quantitative accuracy. The Symmlet filter with vanishing moment of 4 and scale parameter of 20 is used to improve the smoothness of the spectrum and the stability of modeling under the premise of ensuring the resolution. The high-resolution spectrum obtained after wavelet transform is shown in Figure 4 .

[0059] In order to remove the interference of redundant information of the spectrum on modeling, the competitive adaptive reweighted sampling method is used to screen the spectral wavelength variables, and the characteristic wavelengths highly related to urea concentration or uremia state are selected for modeling.

[0060] First, the partial least squares modeling is performed on the spectral data containing all wavelengths, and the regression coefficient of each wavelength variable for the model is obtained, wherein the higher the absolute value of the correlation coefficient, the more beneficial to modeling; then, according to the absolute value of the regression coefficient, each wavelength is sorted, and the wavelengths with relatively small distance are deleted by using an exponential decay function, i.e. the i second screening, the retention rate of the number of wavelengths is determined according to the formula ; wherein, is the retention rate of the number of wavelengths in the i second screening, and are constants, , , M is the total number of initial wavelengths, N is the number of repeated screening cycles. For the modeling model after each screening, the root mean square error is calculated by cross-validation, and after N screening cycles, the combination of wavelength variables with the smallest cross-validation root mean square error is finally selected.

[0061] 4) Based on the screened spectral wavelength variables, a urea concentration quantitative regression model and a uremia state discriminant classification model are constructed.

[0062] Specifically, the partial least squares regression and cross-validation technology are used, the important wavelengths are screened by using the competitive adaptive reweighted sampling method, the preprocessed spectral data in the training set are modeled, then the established model is used to analyze the serum samples in the test set, and the urea concentration is predicted.

[0063] The input data of the urea concentration quantitative regression model is the preprocessed spectral matrix X and the urea concentration matrix ; wherein, , is the absorption intensity value of the spectrum of the n th serum sample at the m th wave number, , is the urea concentration of the n th serum sample, n is the total number of serum samples, m is the total number of wave numbers.

[0064] The input data of the uremia state discriminant classification model is the preprocessed spectral matrix X and the uremia state matrix ; wherein, , is the uremia state of the n th serum sample.

[0065] 5) Using the established model to predict the test set, and combining the real test results to evaluate the model performance.

[0066] For urea concentration quantitative regression model, by comparing the difference between the predicted value and the test reference value, the root mean square error and the determination coefficient R 2 =1- SS res / SS tot are used to evaluate the accuracy of urea concentration quantitative regression model. Among them, n is the number of serum samples, is the reference concentration value of urea index on the test sheet of the j th serum sample, is the predicted value of urea concentration obtained by using urea concentration quantitative regression model to analyze the j th serum sample, SS res is the sum of squares of the difference between the observed value and the predicted value, SS tot is the sum of squares of the difference between the observed value and the average of the observed value.

[0067] The results of urea concentration quantitative regression model are shown in Table 1. Based on the spectral data processed by continuous wavelet transform, the urea concentration quantitative regression model established in the wave number range of 4900~4200cm -1 has the best effect. This waveband covers the important absorption region of urea molecules and related nitrogen-containing compounds, and has good representativeness and specificity. Under this pretreatment and waveband selection scheme, the prediction performance of the partial least squares regression model on the validation set and the test set is the best, and the corresponding cross-validation root mean square error and prediction standard deviation are 1.29 and 0.72 respectively, which is significantly better than the modeling results of other spectral pretreatment methods. As shown in Figure 5 and Figure 6 , the trend of root mean square error of the model with the number of latent variables (number of factors) shows that when the number of factors reaches 7, the performance of the model tends to be stable, and further increasing the number of factors has limited effect on improving the accuracy of the model, and may lead to overfitting. Under the condition of the optimal number of factors, the prediction results of the model and the actual reference values show a high consistency, and the determination coefficient between the predicted value of urea concentration and the test reference value reaches 0.998, indicating that the model has high prediction accuracy and robustness.

[0068] Table 1 Results of urea concentration quantitative regression model

[0069]

[0070] For uremia state discriminant classification model, by comparing the model predicted value with the test diagnosis information, the validation set Q2 Y and the accuracy, precision and recall of the test set are used to evaluate the classification effect of the uremia state discrimination classification model.

[0071] The accuracy is the proportion of the number of samples correctly predicted by the model to the total number of samples, and the calculation formula is ; wherein TP is true positive, TN is true negative, FP is false positive, and FN is false negative.

[0072] The precision is the proportion of the actual positive samples in the samples predicted as positive, and the calculation formula is .

[0073] The recall is the proportion of the samples successfully predicted as positive in the actual positive samples, and the calculation formula is .

[0074] is calculated by cross-validation, which is used to evaluate the prediction ability of the model, and the calculation formula is ; wherein is the uremia state of the urea index on the test sheet of the serum sample, j is the uremia state prediction value obtained by analyzing the serum sample using the uremia state discrimination classification model, is the average value of the reference value of the validation set sample. j The value of is closer to 1, indicating that the prediction effect of the model is better. The modeling results of the uremia state discrimination classification model are shown in Table 2. In this embodiment, the original spectrum with a wave number range of 6500-5700 cm -1 was selected as the input variable, and the competitive adaptive reweighted sampling method was used to select the key wavelength variable to construct the uremia state discrimination classification model. This scheme has the best performance in model performance. The discrimination model constructed on the basis of this variable selection has excellent classification ability, and the prediction performance index

[0075] 2 Q Y up to 0.99, indicating that the uremia state discrimination classification model established under this condition has the highest discrimination degree between the two types of samples, and the classification result accuracy of the test set reaches 100%, realizing the complete correct discrimination of uremia positive and negative samples, indicating that the model has strong discrimination ability and clinical applicability. For example, Figure 7 and Figure 8 ​​As shown, with the increase of the number of model factors, the root mean square error gradually decreases and tends to be stable when the number of factors is 6, indicating that the model has both accuracy and robustness at this time. Under the condition of the optimal factor number, the prediction results of the model on the validation set and the test set samples show that the positive (uremia) samples and the negative (healthy) samples are well distinguished, the classification boundary is clear, and there is no cross or misjudgment, which further verifies the practical value and promotion potential of the model.

[0076] Table 2 Uremia state discrimination classification model results

[0077]

[0078] The present application can realize accurate determination of urea concentration and rapid automatic identification of uremia, and the spectral measurement time is less than 1 minute and the data processing time is less than 0.5 seconds. Specifically, first, the spectral resolution is improved by continuous wavelet transform, and the identifiable of the characteristic waveband of biomolecules in serum samples is enhanced; second, a variable selection algorithm is used to select important wavelengths significantly related to urea concentration and uremia from the full spectrum; further, different spectral pretreatment and modeling strategies are compared, and model parameters are optimized, and high-accuracy regression model and classification model are constructed respectively to realize quantitative prediction of urea concentration in serum and intelligent discrimination of uremia. The whole process is simple, fast and stable, and can provide reliable data support for early screening and diagnosis of clinical uremia, and has good clinical application prospect.

[0079] The present application only needs to collect the near-infrared transmission spectrum data of the serum sample to be tested, so as to accurately detect the urea concentration and judge whether there is a risk of uremia without relying on chemical reagents and complex operations. Compared with the traditional method of jointly judging multiple clinical indicators, the present application has significant advantages in reducing cost, simplifying process and improving diagnosis efficiency, and can provide a fast, intelligent and accurate data support platform for clinical kidney function screening and early identification of uremia.

[0080] It should be noted that the user information (including but not limited to user equipment information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in the present application are all information and data authorized by the user or authorized by all parties, and the collection, use and processing of related data need to comply with relevant regulations.

[0081] In the present application, all actions of obtaining signals, information or data are carried out on the premise of complying with the corresponding data protection regulations and policies of the place, and obtaining the authorization given by the owner of the corresponding device.

[0082] Any technical features in the above embodiments can be combined, and for the sake of brevity, not all possible combinations are described above, however, it should be understood that the application encompasses all possible combinations of the technical features unless such a combination is not technically possible.

[0083] The principles and implementation manners of the present application are described herein by using specific examples, and the above embodiments are only used to help understand the method of the present application and its core idea; meanwhile, for those skilled in the art, the specific implementation manners and application range can be changed according to the idea of the present application. In conclusion, the content of the present specification should not be understood as a limitation of the present application.

Claims

1. A quantitative-discriminative integrated uremia intelligent detection system, characterized in that, The system comprises: a spectrum acquisition module, configured to acquire near-infrared transmission spectrum data of a serum sample to be measured; a double-model detection module, configured to determine urea concentration and uremia state of the serum sample to be measured by using a pre-constructed urea concentration quantitative regression model and a uremia state discriminant classification model respectively according to the near-infrared transmission spectrum data. The urea concentration quantitative regression model is constructed by using a partial least squares method according to a first sample set in advance; the uremia state discriminant classification model is constructed by using a partial least squares method according to a second sample set in advance; the first sample set comprises near-infrared transmission spectrum data of multiple serum samples and urea concentration of each serum sample; and the second sample set comprises near-infrared transmission spectrum data of multiple serum samples and uremia state of each serum sample.

2. The quantitative-discriminative integrated uremia intelligent detection system according to claim 1, wherein, The spectrum acquisition module acquires the near-infrared transmission spectrum data of the serum sample to be measured under constant temperature; the serum sample to be measured is located in a cuvette, and the cuvette is placed in a constant-temperature sample holder.

3. The integrated quantification and discrimination system for uremia according to claim 1, wherein The double-model detection module comprises: a spectrum processing submodule, configured to generate model input data according to the near-infrared transmission spectrum data; the model input data comprises absorption intensity values of the near-infrared transmission spectrum data at each key characteristic wavelength; a quantitative detection submodule, configured to determine urea concentration of the serum sample to be measured by using the pre-constructed urea concentration quantitative regression model according to the model input data; a state discriminant submodule, configured to determine uremia state of the serum sample to be measured by using the pre-constructed uremia state discriminant classification model according to the model input data.

4. The integrated quantification and discrimination system for uremia according to claim 3, wherein The spectrum processing submodule comprises: a spectrum transformation unit, configured to perform continuous wavelet transformation on the near-infrared transmission spectrum data; a wavelength screening unit, configured to perform wavelength screening on the near-infrared transmission spectrum data after continuous wavelet transformation by using a competitive adaptive reweighted sampling method to obtain key characteristic wavelengths; an input data construction submodule, configured to determine model input data according to the key characteristic wavelengths and the near-infrared transmission spectrum data.

5. The integrated quantification and discrimination system for uremia according to claim 4, wherein The spectrum transformation unit performs continuous wavelet transformation on the near-infrared transmission spectrum data by using a Symmlet filter with a vanishing moment of 4.

6. The integrated quantifying and discriminating system for uremia according to claim 1, wherein, The multiple serum samples comprise serum samples of normal people and serum samples of uremia patients.

7. The integrated quantifying and discriminating system for uremia according to claim 1, wherein, The uremia state is 1 or -1, 1 representing positive and -1 representing negative.

8. The integrated quantifying and discriminating system for uremia according to claim 1, wherein, The spectrum acquisition module and the double-model detection module are both arranged in a computer device.

9. The integrated quantifying and discriminating system for uremia according to claim 1, wherein, The system further comprises: a visualization module, configured to display urea concentration and uremia state of the serum sample to be measured to assist doctors in diagnosing uremia.

Images