Klebsiella pneumoniae carbapenemase subtype detection system and method based on CNN (Convolutional Neural Network) and matrix-assisted laser desorption ionization time-of-flight mass spectrometry and application

Abstract

The invention discloses a Klebsiella pneumoniae carbapenemase subtype detection system and method based on CNN and matrix-assisted laser desorption / ionization time-of-flight mass spectrometry (MALDI-TOF MS) and application, the system comprises: a sample processing module for processing a to-be-detected strain to obtain a test sample; the data acquisition module is used for acquiring mass spectrum data of the test sample by using a mass spectrometer; and the CNN identification model is constructed on the basis of a CNN network and is used for analyzing the mass spectrum data of the test sample to obtain the variety of the carbapenem subtype in the strain to be detected. The Klebsiella pneumoniae carbapenemase subtype detection system and the detection method are provided on the basis of a matrix-assisted laser desorption ionization time-of-flight mass spectrometry technology in combination with an artificial intelligence convolutional neural network, and rapid detection of the Klebsiella pneumoniae carbapenemase subtype can be realized; the kit can provide a basis for clinical diagnosis and precise treatment of klebsiella pneumoniae, and has a wide application prospect.

Description

Technical Field

[0001] The present invention relates to the field of biological detection technology, and in particular to a Klebsiella pneumoniae carbapenemase subtype detection system, method and application based on CNN and matrix-assisted laser desorption ionization time-of-flight mass spectrometry. Background Art

[0002] Carbapenemase-resistant Klebsiella pneumoniae (CRKP) is a common opportunistic Gram-negative bacteria in hospitals, which can cause high morbidity and mortality. One of the main resistance mechanisms of CRKP is the production of carbapenemases. Different types of carbapenemases have different infection-related mortality rates and responses to the same drug. Therefore, the rapid detection and classification of carbapenemases are crucial for the timely treatment and control of infection. The existing enzyme type detection methods are either time-consuming, require special equipment, or are expensive, which cannot meet clinical needs. Matrix-assisted laser desorption ionization time-of-flight mass spectrometry (MALDI-TOF MS) can quickly identify species based on only a few features such as m / z and peak height. It has the advantages of time-saving, sensitivity, simplicity, reliability, and strong applicability. It has been widely used in microbiology laboratories in recent years. However, bacterial resistant enzymes usually have many subtypes, and their mass spectra have high similarity, which makes it difficult to accurately distinguish them. Therefore, there is an urgent need for a technical solution that can accurately and quickly detect Klebsiella pneumoniae carbapenemase subtypes. Summary of the invention

[0003] The technical problem to be solved by the present invention is to provide a Klebsiella pneumoniae carbapenemase subtype detection system, method and application based on CNN and matrix-assisted laser desorption ionization time-of-flight mass spectrometry (MALDI-TOF MS) in view of the deficiencies in the above-mentioned prior art.

[0004] In order to solve the above technical problems, the technical solution adopted by the present invention is: In a first aspect of the present invention, a Klebsiella pneumoniae carbapenemase subtype detection system based on CNN and matrix-assisted laser desorption ionization time-of-flight mass spectrometry is provided, comprising: A sample processing module, which processes the strain to be tested to obtain a test sample; A data acquisition module, using a mass spectrometer to collect mass spectrum data of the test sample; The CNN identification model is constructed based on the CNN network and is used to obtain the type of carbapenem subtype in the test strain based on the mass spectrometry data of the test sample.

[0005] Preferably, the sample processing module uses a formic acid extraction method to obtain the supernatant of the strain to be tested as a test sample.

[0006] Preferably, the method for the sample processing module to obtain the supernatant of the strain to be tested is: 1) Use an inoculating loop to pick up several medium-sized fresh colonies on the blood plate after the second subculture and put them into a centrifuge tube, add sterile water to mix, and then add anhydrous ethanol and vortex to mix; 2) centrifuging the solution obtained in step 1), discarding the supernatant and centrifuging again to remove the residual supernatant, and drying the precipitate at room temperature; 3) adding formic acid solution to the product obtained in step 2) and mixing well, and leaving at room temperature; 4) Add acetonitrile to the product obtained in step 3), mix thoroughly and centrifuge to obtain the supernatant of the strain to be tested.

[0007] Preferably, the method for the sample processing module to obtain the supernatant of the strain to be tested is: 1) Use an inoculation loop to pick 2-3 medium fresh colonies on the blood plate after the second subculture and put them into a 1.5 mL centrifuge tube, add 200 μL of sterile water to mix, and then add 600 μL of anhydrous ethanol and vortex to mix; 2) Centrifuge the solution obtained in step 1) at 12000×g for 2 min, discard the supernatant and centrifuge again for 1 min, remove the residual supernatant, and dry the precipitate at room temperature for 5 min; 3) Add 80 μL of 70 wt% formic acid solution to the product obtained in step 2), mix well, and leave at room temperature for 5 min; 4) Add 80 μL of acetonitrile to the product obtained in step 3), mix thoroughly, and centrifuge at 12,000 × g for 3 min to obtain the supernatant of the strain to be tested.

[0008] Preferably, the method by which the data acquisition module acquires mass spectrum data of the test sample is: The test sample is dripped onto the orifice plate, and after drying, the matrix is ​​dripped to cover the test sample, and then the mass spectrometer is used to collect the mass spectrum data of the test sample.

[0009] Preferably, the CNN identification model is constructed by the following method: S1. Build training data set: S1-1, incubate Klebsiella pneumoniae on a blood plate, detect the bacterial species and isolate the pathogenic bacteria as a test sample; S1-2, extracting the test sample obtained in step S1-1 by formic acid to obtain the supernatant of the strain to be tested, and then using the data acquisition module to collect mass spectrometry data of the supernatant of the strain to be tested; S1-3, obtaining mass spectrometry data of different Klebsiella pneumoniae carbapenemase subtypes through step S1-1 and step S1-2, and setting a label for the mass spectrometry data of each subtype, wherein the label is the corresponding bacterial subtype; S1-4, smoothing, baseline correction, and normalization are performed on the mass spectrum of each subtype of bacterial species to obtain the data of the subtype of bacterial species, and the mass spectrum data of all subtypes of bacterial species are combined to obtain a training data set; S2. Use the CNN network as the basic model, take the mass spectrometry data of the strain as input, and the label corresponding to the strain as output, use the cross entropy function as the loss function, and use the training data set to train the CNN network. After the training is completed, the CNN identification model is obtained.

[0010] Preferably, the CNN identification model includes an input layer, three pre-processing units, a flattening layer, two fully connected layers and an output layer connected in sequence.

[0011] Preferably, each pre-processing unit includes a first convolutional layer, a first batch of normalized layers, a second convolutional layer, a second batch of normalized layers and a maximum pooling layer connected in sequence, and a Relu activation function is connected between the first batch of normalized layers and the second convolutional layer, and between the second batch of normalized layers and the maximum pooling layer; There is a Relu activation function connected between the two fully connected layers, and a Softmax function is connected between the last fully connected layer and the output layer; The convolution kernel sizes of the first and second convolutional layers are both 3*1, and the sliding window size of the maximum pooling layer is 3*1.

[0012] The second aspect of the present invention provides a method for detecting Klebsiella pneumoniae carbapenemase subtypes based on CNN and matrix-assisted laser desorption ionization time-of-flight mass spectrometry, characterized in that the system as described above is used to identify Klebsiella pneumoniae carbapenemase subtypes, and the method comprises the following steps: Step 1: obtaining the supernatant of the strain to be tested as a test sample through a sample processing module; Step 2: Collecting mass spectrum data of the test sample through a data acquisition module; Step 3: Input the mass spectrometry data of the test sample into the CNN identification model, and the CNN identification model outputs the identification results of the Klebsiella pneumoniae carbapenemase subtypes.

[0013] The third aspect of the present invention provides an application of the system or method as described above in pathogen resistance testing, the application method being: after the carbapenemase subtype of Klebsiella pneumoniae is identified by the system or method, a drug sensitivity analyzer is used to detect the drug resistance of the identified pathogen.

[0014] The beneficial effects of the present invention are: The present invention is based on matrix-assisted laser desorption ionization time-of-flight mass spectrometry technology, combined with artificial intelligence convolutional neural network, and provides a Klebsiella pneumoniae carbapenemase subtype detection system and detection method based on CNN and matrix-assisted laser desorption ionization time-of-flight mass spectrometry. The present invention can achieve rapid detection of Klebsiella pneumoniae carbapenemase subtypes, and can provide a basis for clinical diagnosis and precise treatment of Klebsiella pneumoniae. The detection scheme of the present invention is simple, easy to operate, fast, and has broad application prospects. The scheme of the present invention can also be further applied in pathogen resistance testing. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 It is a structural diagram of the CNN identification model of Example 1; Figure 2 A flow chart for obtaining a mass spectrum in Example 1; Figure 3 ROC curves for identifying three Klebsiella pneumoniae carbapenemase subtypes using four machine learning schemes in Example 1; Figure 4 This is the confusion matrix of mass spectra of CNN identifying three Klebsiella pneumoniae carbapenemase subtypes in Example 1. DETAILED DESCRIPTION

[0016] The present invention is further described in detail below in conjunction with embodiments so that those skilled in the art can implement the invention with reference to the description.

[0017] It should be understood that the terms such as “having”, “including” and “comprising” used herein do not exclude the existence or addition of one or more other elements or combinations thereof.

[0018] The test methods used in the following examples are conventional methods unless otherwise specified. The materials and reagents used in the following examples are all commercially available unless otherwise specified. In the following examples, if no specific conditions are specified, the experiments were carried out under conventional conditions or conditions recommended by the manufacturer. The reagents or instruments used, if the manufacturer is not specified, are all conventional products that can be purchased commercially. Example

[0019] A Klebsiella pneumoniae carbapenemase subtype detection system based on CNN and matrix-assisted laser desorption ionization time-of-flight mass spectrometry, comprising: A sample processing module, which processes the strain to be tested to obtain a test sample; A data acquisition module, using a mass spectrometer to collect mass spectrum data of the test sample; The CNN identification model is constructed based on the CNN network and is used to obtain the type of carbapenem subtype in the test strain based on the mass spectrometry data of the test sample.

[0020] In this embodiment, the sample processing module uses formic acid extraction to obtain the supernatant of the strain to be tested as a test sample.

[0021] The supernatant of the strain to be tested was prepared by the following method: 1) Use an inoculation loop to pick 2-3 medium fresh colonies on the blood plate after the second subculture and put them into a 1.5 mL centrifuge tube, add 200 μL of sterile water to mix, and then add 600 μL of anhydrous ethanol and vortex to mix; 2) Centrifuge the solution obtained in step 1) at 12000×g for 2 min, discard the supernatant and centrifuge again for 1 min, remove the residual supernatant, and dry the precipitate at room temperature for 5 min; 3) Add 80 μL of 70 wt% formic acid solution to the product obtained in step 2), mix well, and leave at room temperature for 5 min; 4) Add 80 μL of acetonitrile to the product obtained in step 3), mix thoroughly, and centrifuge at 12,000 × g for 3 min to obtain the supernatant of the strain to be tested.

[0022] In this embodiment, the method in which the data acquisition module acquires mass spectrum data of the test sample is: The test sample is dripped onto a 96-well plate, and after drying, the matrix is ​​dripped onto the test sample to cover the test sample, and then the mass spectrometer is used to collect mass spectrum data of the test sample.

[0023] In this embodiment, the CNN identification model is constructed by the following method: S1. Build training data set: S1-1. Incubate the pathogens on a blood plate, detect the strains and separate the pathogens as test samples. The specific steps are as follows: (1) Collection objects Methods: Klebsiella pneumoniae isolates were collected from the Microbiology Laboratory of the Laboratory Department of the Affiliated Hospital of Xuzhou Medical University from 2018 to 2023; the collected Klebsiella pneumoniae were transferred to blood agar plates and cultured aerobically at 35 ℃ for 24 h, and single colonies were selected for secondary passage (culture conditions were the same as before), and the strains were confirmed by the direct smear method using a Zhongyuan Huiji mass spectrometer, and Klebsiella pneumoniae with a score ≥2.0 were included in the analysis.

[0024] (2) Phenotypic judgment The single pure colony isolated from the blood plate was verified by mass spectrometry, and the drug sensitivity test was performed by VITEK-2 Compact. The result interpretation was based on the breakpoints recommended by CLSI 2022 M100 34th edition; disk diffusion method (KB method) drug sensitivity test: imipenem or meropenem or doripenem inhibition ring ≥ 23 mm is sensitive strain (CSKP), < 19 mm is resistant strain (CRKP). ATCC BAA-1705 is a quality control strain for antimicrobial drug sensitivity test and gene detection.

[0025] (3) Genotype testing Polymerase chain reaction (PCR) was used to preliminarily screen the carbapenemase genes of the strains.

[0026] S1-2, extracting the test sample obtained in step S1-1 by formic acid to obtain the supernatant of the strain to be tested, and then using a data acquisition module to collect the mass spectrum of the sample to be tested; Extract supernatant (1.1) Use an inoculation loop to pick 2-3 medium fresh colonies on the blood plate after the second subculture and place them in a 1.5 ml centrifuge tube. Add 200 μl of sterile water and mix well. Add 600 μl of anhydrous ethanol and vortex to mix well. (1.2) The solution in step (1.1) was centrifuged at 12000×g for 2 min, the supernatant was discarded and centrifuged again for 1 min, the residual supernatant was removed, and the precipitate was dried at room temperature for 5 min; (1.3) Add 80 μl of 70 wt% formic acid to the product obtained in step (1.2), mix well, and leave at room temperature for 5 min; (1.4) Add 80 μl of acetonitrile to the product obtained in step (1.3), mix thoroughly, and centrifuge at 12,000 × g for 3 min to obtain the supernatant of the strain to be tested.

[0027] Mass spectrometry acquisition Take 1 μl of the supernatant of the sample to be tested and drop it on a 96-well plate. After drying, add 1 μl of matrix to cover the sample, and then use Zhongyuan Huiji mass spectrometer (MALDI-TOF MS) to collect the mass spectrum of the sample to be tested. 12 samples are spotted for each bacterial strain, and each target is hit once, that is, 12 spectra are generated for each bacterial strain. After eliminating the spectra with a score of <2.0, the obtained spectra are sorted for use. The flow chart is as follows Figure 2 shown.

[0028] S1-3. Obtain mass spectrometry data of different Klebsiella pneumoniae carbapenemase subtypes through step S1-1 and step S1-2, and set a label for the mass spectrometry data of each subtype, where the label is the corresponding strain subtype. The mass spectrometry data of the collected strains are shown in Table 1; Table 1 type Number of plants Number of spectra (number of strains * number of spectra per strain) IMP 1 12（1*12） NDM 47 564（47*12） KPC 111 1332（111*12） OXA 47 564（47*12） total 206 2472 S1-4, smoothing, baseline correction, and normalization are performed on the mass spectrum of each subtype of bacterial species to obtain the data of the subtype of bacterial species, and the mass spectrum data of all subtypes of bacterial species are combined to obtain a training data set; S2. Use the CNN network as the basic model, take the mass spectrometry data of the strain as input and the label corresponding to the strain as output, use the cross entropy function as the loss function, use the training data set to train the CNN network, and obtain the CNN identification model after training.

[0029] Reference Figure 1 ,In this embodiment, the CNN identification model includes an input layer (Input), three pre-processing units, a flat layer (Flatten), two fully connected layers (Linear) and an output layer (Output) that are connected in sequence; Each pre-processing unit includes the first convolution layer (Conv1d), the first batch normalization layer (BatchNorm1d), the second convolution layer (Conv1d), the second batch normalization layer (BatchNorm1d) and the maximum pooling layer (MaxPool1d) connected in sequence, and the Relu activation function is connected between the first batch normalization layer and the second convolution layer, and between the second batch normalization layer and the maximum pooling layer; There is a Relu activation function connected between the two fully connected layers, and a Softmax function is connected between the last fully connected layer and the output layer; The convolution kernel sizes of the first and second convolutional layers are both 3*1, and the sliding window size of the maximum pooling layer is 3*1.

[0030] Among them, configuring a batch normalization layer and a Relu activation function after each convolutional layer can improve the training speed, enhance the network's expressiveness, and reduce overfitting; the addition of the pooling layer enhances the model's ability to learn local and global features; the flattening layer converts the multi-dimensional feature representation into a one-dimensional form to input the features output by the pooling layer into the fully connected layer; finally, the fully connected layer maps these features to the output category.

[0031] In this example, in order to evaluate the classification and prediction capabilities of the system (CNN) of the present invention for matrix-assisted laser desorption ionization time-of-flight mass spectrometry (MALDI-TOF MS) spectra, three traditional machine learning algorithms were selected for comparison with CNN, namely, random forest (RF), support vector machine (SVM), and Adaboost. These algorithms are deployed based on Scikit-learn and use default parameter configurations.

[0032] In this embodiment, the accuracy rate (ACC), AUC (Area Under the Curve), precision rate (Pre), recall rate (Recall) and F1 score are selected to score several algorithms. Accuracy rate (ACC) refers to the proportion of correct results predicted by the model, precision rate (Pre) refers to the proportion of samples that actually belong to the positive class among the samples predicted to be positive, recall rate (Recall) refers to the proportion of predicted positive classes to actual positive classes, and F1 score combines the results of Precision and Recall to evaluate the effect of the classification model.

[0033] In this embodiment, a five-fold cross validation was used to evaluate the performance of several machine learning models. The original mass spectrometry data was randomly and evenly divided into five subsets, four of which were selected as training sets and one subset as a test set. The model was trained and evaluated on the test set. This process was repeated five times and the average value was taken to obtain the final evaluation result.

[0034] The results of the comparison of the classification and prediction capabilities of the four machine learning methods for the three isolated Klebsiella pneumoniae carbapenemase subtypes are shown in Table 2: Table 2 Algorithm Accuracy AUC Precision Recall F1 RF 0.9289 0.9917 0.9312 0.9289 0.9270 SVM 0.8394 0.9558 0.8658 0.8394 0.8254 AdaBoost 0.9126 0.9656 0.9150 0.9126 0.9106 CNN 0.9606 0.9910 0.9610 0.9606 0.9603 Reference Figure 3 ROC curves for identification of three Klebsiella pneumoniae carbapenemase subtypes for four machine learning algorithms.

[0035] Reference Figure 4 Confusion matrix of mass spectra for CNN identification of three Klebsiella pneumoniae carbapenemase subtypes.

[0036] It can be seen from the test results that the CNN identification model of this embodiment has obvious advantages in accuracy, precision, recall rate, F1 score and other dimensions compared with several other traditional machine learning algorithm models.

[0037] Although the embodiments of the present invention have been disclosed as above, they are not limited to the applications listed in the specification and the implementation modes. They can be fully applied to various fields suitable for the present invention. For those familiar with the art, additional modifications can be easily implemented. Therefore, without departing from the general concept defined by the claims and the scope of equivalents, the present invention is not limited to specific details.

Claims

1. A Klebsiella pneumoniae carbapenemase subtype detection system based on CNN and matrix-assisted laser desorption ionization time-of-flight mass spectrometry, characterized in that: include: A sample processing module, which processes the strain to be tested to obtain a test sample; A data acquisition module, using a mass spectrometer to collect mass spectrum data of the test sample; The CNN identification model is constructed based on the CNN network and is used to obtain the type of carbapenem subtype in the test strain based on the mass spectrometry data of the test sample.

2. The Klebsiella pneumoniae carbapenemase subtype detection system based on CNN and matrix-assisted laser desorption ionization time-of-flight mass spectrometry according to claim 1, characterized in that: The sample processing module uses a formic acid extraction method to obtain the supernatant of the strain to be tested as a test sample.

3. The Klebsiella pneumoniae carbapenemase subtype detection system based on CNN and matrix-assisted laser desorption ionization time-of-flight mass spectrometry according to claim 2, characterized in that: The method for the sample processing module to obtain the supernatant of the strain to be tested is: 1) Use an inoculating loop to pick up several medium-sized fresh colonies on the blood plate after the second subculture and put them into a centrifuge tube, add sterile water to mix, and then add anhydrous ethanol and vortex to mix; 2) centrifuging the solution obtained in step 1), discarding the supernatant and centrifuging again to remove the residual supernatant, and drying the precipitate at room temperature; 3) adding formic acid solution to the product obtained in step 2) and mixing well, and leaving at room temperature; 4) Add acetonitrile to the product obtained in step 3), mix thoroughly and centrifuge to obtain the supernatant of the strain to be tested.

4. The Klebsiella pneumoniae carbapenemase subtype detection system based on CNN and matrix-assisted laser desorption ionization time-of-flight mass spectrometry according to claim 3, characterized in that: The method for the sample processing module to obtain the supernatant of the strain to be tested is: 1) Use an inoculation loop to pick 2-3 medium fresh colonies on the blood plate after the second subculture and put them into a 1.5 mL centrifuge tube, add 200 μL of sterile water to mix, and then add 600 μL of anhydrous ethanol and vortex to mix; 2) Centrifuge the solution obtained in step 1) at 12000×g for 2 min, discard the supernatant and centrifuge again for 1 min, remove the residual supernatant, and dry the precipitate at room temperature for 5 min; 3) Add 80 μL of 70 wt% formic acid solution to the product obtained in step 2), mix well, and leave at room temperature for 5 min; 4) Add 80 μL of acetonitrile to the product obtained in step 3), mix thoroughly, and centrifuge at 12,000 × g for 3 min to obtain the supernatant of the strain to be tested.

5. The Klebsiella pneumoniae carbapenemase subtype detection system based on CNN and matrix-assisted laser desorption ionization time-of-flight mass spectrometry according to claim 1, characterized in that: The method for the data acquisition module to acquire mass spectrum data of the test sample is: The test sample is dripped onto the orifice plate, and after drying, the matrix is ​​dripped to cover the test sample, and then the mass spectrometer is used to collect the mass spectrum data of the test sample.

6. The Klebsiella pneumoniae carbapenemase subtype detection system based on CNN and matrix-assisted laser desorption ionization time-of-flight mass spectrometry according to claim 1, characterized in that: The CNN identification model is constructed by the following method: S1. Build training data set: S1-1, incubate Klebsiella pneumoniae on a blood plate, detect the bacterial species and isolate the pathogenic bacteria as a test sample; S1-2, extracting the test sample obtained in step S1-1 by formic acid to obtain the supernatant of the strain to be tested, and then using the data acquisition module to collect mass spectrometry data of the supernatant of the strain to be tested; S1-3, obtaining mass spectrometry data of different Klebsiella pneumoniae carbapenemase subtypes through step S1-1 and step S1-2, and setting a label for the mass spectrometry data of each subtype, wherein the label is the corresponding bacterial subtype; S1-4, smoothing, baseline correction, and normalization are performed on the mass spectrum of each subtype of bacterial species to obtain the data of the subtype of bacterial species, and the mass spectrum data of all subtypes of bacterial species are combined to obtain a training data set; S2. Use the CNN network as the basic model, take the mass spectrometry data of the strain as input, and the label corresponding to the strain as output, use the cross entropy function as the loss function, and use the training data set to train the CNN network. After the training is completed, the CNN identification model is obtained.

7. The Klebsiella pneumoniae carbapenemase subtype detection system based on CNN and matrix-assisted laser desorption ionization time-of-flight mass spectrometry according to claim 6, characterized in that: The CNN identification model includes an input layer, three pre-processing units, a flattening layer, two fully connected layers and an output layer connected in sequence.

8. The Klebsiella pneumoniae carbapenemase subtype detection system based on CNN and matrix-assisted laser desorption ionization time-of-flight mass spectrometry according to claim 7, characterized in that: Each pre-processing unit includes a first convolution layer, a first normalization layer, a second convolution layer, a second normalization layer and a maximum pooling layer connected in sequence, and a Relu activation function is connected between the first normalization layer and the second convolution layer, and between the second normalization layer and the maximum pooling layer; There is a Relu activation function connected between the two fully connected layers, and a Softmax function is connected between the last fully connected layer and the output layer; The convolution kernel sizes of the first and second convolutional layers are both 3*1, and the sliding window size of the maximum pooling layer is 3*1.

9. A method for detecting Klebsiella pneumoniae carbapenemase subtypes based on CNN and matrix-assisted laser desorption ionization time-of-flight mass spectrometry, characterized in that: The method uses the system according to any one of claims 1 to 6 to identify the Klebsiella pneumoniae carbapenemase subtype, and the method comprises the following steps: Step 1: obtaining the supernatant of the strain to be tested as a test sample through a sample processing module; Step 2: Collecting mass spectrum data of the test sample through a data acquisition module; Step 3: Input the mass spectrometry data of the test sample into the CNN identification model, and the CNN identification model outputs the identification results of the Klebsiella pneumoniae carbapenemase subtypes.

10. Use of the system according to any one of claims 1 to 8 or the method according to claim 9 in testing the drug resistance of pathogens, characterized in that: The application method is: after the carbapenemase subtype of Klebsiella pneumoniae is identified by the system or method, a drug sensitivity analyzer is used to detect the drug resistance of the identified pathogenic bacteria.

Images