Bacterial drug resistance detection method and system

Abstract

The invention discloses a bacterial drug resistance detection method and system, and particularly relates to the technical field of bacterial drug resistance detection.The bacterial drug resistance detection method comprises the following steps that firstly, a bacterial infection sample is collected and subjected to primary treatment, impurities are removed, bacteria are separated and inoculated to a pre-prepared culture medium, and a single strain is cultured and purified in a preset environment; then, distributing a single strain sample into reaction holes with different drug concentrations by using automatic equipment, recording expression data of strains in each hole, and generating an expression data set; performing feature extraction and analysis based on the data set, extracting key features such as a bacterial growth inhibition index and a drug resistance fluctuation index, inputting an analysis result into a pre-trained machine learning model, and finally outputting a drug resistance grade of a single strain; according to the method, the drug resistance characteristics of the bacteria can be comprehensively and accurately captured through the bacterial growth inhibition index and the drug resistance expression fluctuation index, and the drug resistance level of the bacteria can be more scientifically and comprehensively evaluated through combination of the two indexes.

Description

Technical Field

[0001] The present invention relates to the technical field of bacterial drug resistance detection, and more specifically, to a bacterial drug resistance detection method and system. Background Art

[0002] Bacterial resistance testing refers to the use of experimental methods to detect the sensitivity or resistance of bacteria to specific antibiotics in order to determine the therapeutic effect of antibiotics on bacterial infections. This process involves exposing bacterial samples to different concentrations of antibiotics, analyzing bacterial growth, and determining their resistance levels, thereby providing a basis for targeted drug selection in the clinic.

[0003] With the widespread use and abuse of antibiotics, bacteria have gradually developed resistance to antimicrobial drugs, resulting in reduced therapeutic efficacy of antimicrobial drugs. Bacterial resistance detection is an important means to solve this problem. Through detection, it is possible to identify which type of antimicrobial drugs bacteria are resistant to, thereby avoiding the use of ineffective drugs and optimizing clinical treatment plans. In addition, the results of resistance detection are of great significance for formulating rational use policies for antimicrobial drugs, monitoring the prevalence trend of resistant strains, and controlling the spread of resistant bacteria. It is an important part of ensuring public health security. Existing methods are mostly based on a single indicator (such as inhibitory concentration), lack a comprehensive evaluation of resistance, and are prone to misjudgment. Therefore, a bacterial resistance detection method and system are proposed in this paper to solve the above problems. Summary of the invention

[0004] To achieve the above object, the present invention provides the following technical solutions:

[0005] A method for detecting bacterial drug resistance comprises the following steps:

[0006] Collect bacterial infection samples and perform preliminary processing on the samples to isolate the existing bacteria and obtain processed samples;

[0007] The processed sample is inoculated onto a pre-configured culture medium and cultured under preset environmental conditions. After obtaining a single bacterial colony, purification culture is performed to ensure that the test object is a single strain;

[0008] Using automated equipment to automatically distribute the test object, i.e., a single strain sample, to reaction wells containing different drug concentrations, and recording the performance data of the single strain sample in the reaction wells with different drug concentrations to obtain a performance data set;

[0009] Feature extraction is performed based on the performance data set, followed by feature analysis. The feature analysis results are input into a pre-trained machine learning model to output the resistance level of a single strain sample.

[0010] In a preferred embodiment, the preliminary treatment refers to: centrifugation, filtration, smearing, and removal of impurities.

[0011] In a preferred embodiment, feature extraction refers to extracting a bacterial growth inhibition feature data group and a drug resistance performance fluctuation feature data group, and then generating a bacterial growth inhibition index based on the bacterial growth inhibition feature data group, and generating a drug resistance performance fluctuation index based on the drug resistance performance fluctuation feature data group.

[0012] In a preferred embodiment, when the detection object, ie, a single strain sample, is automatically distributed to reaction wells containing different drug concentrations, the drug concentrations in adjacent reaction wells change in an isogradient manner and are distributed at the same time.

[0013] In a preferred embodiment, the logic for obtaining the bacterial growth inhibition index is:

[0014] The drug concentration of the i-th reaction well is obtained as C from the bacterial growth inhibition characteristic data set. i , the bacterial growth performance of the i-th reaction well is divided into G i , the score range is [0,1], and then calculate the drug concentration weight W of the i-th reaction well i , n represents the total number of reaction wells, and p represents the drug concentration adjustment coefficient;

[0015] At each drug concentration, the contribution to the inhibition of bacterial growth is introduced with a nonlinear adjustment coefficient λ:

[0016] I i =W i ·(1-G i ) λ ;I i represents the inhibition contribution value of the i-th reaction well;

[0017] Fit a nonlinear regression curve through the concentration-growth performance data:

[0018] k is the slope parameter of the fitting curve, b is the offset parameter of the fitting curve, G(C i ) indicates that the drug concentration is C i When , the fitting score value obtained by fitting the fitting curve;

[0019] Calculate the actual bacterial growth performance score G i and the fitting score G(G i ) to obtain the mean square error E;

[0020] The formula for calculating the bacterial growth inhibition index is:

[0021] n represents the total number of reflecting holes, E max is the preset maximum error value, which is used to normalize the mean square error E. growth It represents the bacterial growth inhibition index.

[0022] In a preferred embodiment, the drug concentration adjustment coefficient p is obtained by the following steps:

[0023] Construct a two-compartment model based on the changes in drug concentration over time in the body:

[0024] C(t)=A·e -αt +B·e -βt ; C(t) represents the concentration of the drug at time t, A and B are constants related to the initial concentration and distribution of the drug, α is the distribution phase rate constant, and β is the elimination phase rate constant. The two-compartment model equation is fitted using the least squares method to solve the values ​​of α and β, which are then substituted into the calculation formula of the drug concentration adjustment coefficient p:

[0025] In a preferred embodiment, the logic for obtaining the drug resistance fluctuation index is:

[0026] Obtain the absolute value of the growth performance difference ΔG between two adjacent drug concentrations from the drug resistance performance fluctuation characteristic data set i , indicating the degree of local fluctuation, based on the absolute value of the difference in growth performance ΔG i time series, calculate the overall volatility standard deviation σ of the time series G , and the average fluctuation amplitude of the time series Considering the impact of drug concentration changes on the degree of fluctuation, the weight between adjacent drug concentrations in the i-th group is defined as W conc，i :

[0027] |C i+1 -C i | represents the absolute difference between adjacent concentrations, indicating the concentration gradient change, j and i are the indexes of the reaction wells, and n represents the total number of reaction wells;

[0028] The calculation formula of drug resistance fluctuation index is:

[0029] I var It represents the fluctuation index of drug resistance.

[0030] In a preferred embodiment, the pre-trained machine learning model is a convolutional neural network model, and the output result of the convolutional neural network model is the drug resistance level of a single strain sample.

[0031] In a preferred embodiment, a bacterial resistance detection system comprises:

[0032] A separation module collects bacterial infection samples and performs preliminary processing on the samples to separate the existing bacteria and obtain processed samples;

[0033] The purification module inoculates the processed sample into a pre-configured culture medium and cultures it under preset environmental conditions. After obtaining a single bacterial colony, it performs purification culture to ensure that the test object is a single strain.

[0034] The detection and recording module uses an automated device to automatically distribute the detection object, i.e., a single strain sample, to reaction wells containing different drug concentrations, and records the performance data of the single strain sample in the reaction wells with different drug concentrations to obtain a performance data set;

[0035] The drug resistance assessment module extracts features based on the performance data set, then performs feature analysis, inputs the feature analysis results into a pre-trained machine learning model, and outputs the drug resistance level of a single strain sample.

[0036] Technical effects and advantages of the present invention:

[0037] The present invention greatly improves the detection efficiency by using automated equipment to distribute and detect bacterial samples. By automatically controlling the distribution of samples in reaction wells with different drug concentrations, drug resistance tests of multiple drug concentrations can be performed at the same time, avoiding errors in traditional manual operations and shortening the detection time. The present invention generates an accurate bacterial growth inhibition index based on the relationship between bacterial growth performance and drug concentration. The bacterial growth inhibition index and the drug resistance performance fluctuation index can comprehensively and accurately capture the drug resistance characteristics of bacteria. The growth inhibition index provides the growth of bacteria under drug intervention, while the drug resistance performance fluctuation index reflects the adaptability and drug resistance performance of bacteria to changes in drug concentration from a time series perspective. The combination of the two can comprehensively evaluate the drug resistance level of bacteria.

[0038] The present invention uses convolutional neural network (CNN) as a machine learning model to perform deep learning on the extracted feature data, so that the bacterial resistance level can be automatically and accurately output. Convolutional neural network can identify complex resistance patterns when processing large-scale data, and has higher accuracy and reliability than traditional rule-based or empirical analysis methods.

[0039] The present invention can more accurately reflect the changing characteristics of drugs in vivo by combining a two-compartment model based on changes in drug concentrations in vivo with a drug concentration adjustment coefficient. This step can further enhance the physiological relevance of the test data and ensure that the obtained drug resistance test results are more in line with clinical practice. This method can simultaneously test bacterial growth under different drug concentrations in multiple reaction wells, has a high throughput capacity, can quickly process multiple samples and multiple drugs, and improves the efficiency of the overall experiment. In addition, the drug resistance of bacteria is not only evaluated by growth inhibition characteristics, but also further analyzed by fluctuations in drug resistance performance, providing multi-dimensional analysis results.

[0040] The present invention calculates the drug concentration weight by introducing a drug concentration adjustment coefficient, thereby enhancing the correlation analysis between drug concentration and bacterial growth inhibition effect, making it more scientific. The nonlinear adjustment parameter and regression fitting curve are introduced to improve the calculation accuracy of the inhibition index and the fluctuation index, providing a reliable basis for subsequent detection models, and avoiding the blind use of antimicrobial drugs by accurately evaluating the drug resistance level, thereby reducing the abuse and indiscriminate use of antibiotics on the environment from the source. The present invention can effectively curb the spread of drug-resistant bacteria, maintain the overall public health and safety of society, provide scientific data support for governments and health organizations, and formulate more effective antimicrobial drug management policies. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to facilitate understanding by those skilled in the art, the present invention is further described below in conjunction with the accompanying drawings;

[0042] Figure 1 The schematic diagram is a schematic diagram of a bacterial resistance detection method in the present invention.

[0043] Figure 2 This is a schematic diagram of a bacterial resistance detection system in the present invention. DETAILED DESCRIPTION

[0044] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0045] Reference Figure 1 - Figure 2 The following embodiments are obtained:

[0046] Embodiment 1: A method for detecting bacterial resistance comprises the following steps:

[0047] Collect bacterial infection samples and perform preliminary processing on the samples to separate the existing bacteria and obtain the processed samples; collecting samples is the first step of the entire test, and the purpose is to obtain samples containing bacteria from patients or the environment. These samples can be blood, urine, sputum, wound secretions, etc. The preliminary processing steps (such as centrifugation, filtration, smear, etc.) are to remove impurities and other non-bacterial substances in the samples, ensure that the samples contain only bacteria, thereby improving the accuracy of the test. Through preliminary processing, bacterial cells can be isolated and prepared for subsequent culture and testing.

[0048] The processed sample is inoculated onto a pre-configured culture medium and cultured under preset environmental conditions. After obtaining a single bacterial colony, it is purified and cultured to ensure that the test object is a single strain. The purpose of inoculating bacteria onto the culture medium and culturing them under appropriate conditions is to allow the bacteria to proliferate in a suitable environment and obtain a sufficient number of bacteria for subsequent experiments. The single colony obtained means that each colony is derived from a single bacterium and reproduced from the same cell. By selecting a single colony, it can be ensured that the bacterial strains used in subsequent tests are of the same variety, avoiding the interference of mixed strains in the test results. The purpose of purification culture is to ensure that there is no interference from other strains during the test process, ensuring that the test object is the target bacteria.

[0049] Use automated equipment to automatically distribute the test object, i.e., a single strain sample, to reaction wells containing different drug concentrations, and record the performance data of the single strain sample in the reaction wells with different drug concentrations to obtain a performance data set; automated equipment can improve experimental efficiency and accurately distribute bacterial samples to reaction wells containing different drug concentrations. The core of this step is to test the response of bacteria to different drug concentrations to evaluate their drug resistance. The drug concentration in the reaction well usually changes in a gradient, so that the inhibition performance of bacteria under different drug concentrations can be determined. By recording the performance data of bacteria at different drug concentrations (such as growth, inhibition, etc.), a data set can be generated to provide a basis for subsequent analysis.

[0050] Feature extraction is performed based on the performance data set, followed by feature analysis. The feature analysis results are input into a pre-trained machine learning model to output the resistance level of a single strain sample. Feature data related to drug resistance are extracted from the performance data set. These features may include bacterial growth inhibition, response to changes in drug concentration, etc. Feature extraction can convert complex data into structures that are easy to analyze and process. Feature analysis: Further analysis of the extracted features to identify patterns or rules related to drug resistance. Machine learning model: Trained machine learning models (such as convolutional neural networks) can automatically evaluate and classify the resistance level of bacteria based on the results of feature analysis. The core of this step is to predict the resistance level of bacteria through the model, thereby determining the resistance of the strain to certain drugs.

[0051] Initial processing refers to: centrifugation, filtration, smearing, and removal of impurities, examples are:

[0052] Sample collection: Use sterile tools to collect samples from the infected site (such as blood, urine, throat swabs, etc.), ensuring that sample contamination is avoided.

[0053] Centrifugation: Place the sample in a centrifuge tube and centrifuge it at an appropriate speed (such as 3000rpm). The purpose is to use centrifugal force to aggregate the bacterial cells in the sediment and separate the cells from the liquid.

[0054] Filtration: If the sample contains larger particles or impurities (such as tissue fragments, cells, etc.), use a filter membrane (such as 0.45μm or 0.2μm filter membrane) to filter it to remove unnecessary impurities and ensure that the filtered liquid is mainly bacteria.

[0055] Smear: Take out the bacterial sample precipitated after centrifugation, take a small amount and evenly spread it on a glass slide to make a bacterial smear for subsequent microscopic observation and staining to preliminarily identify the morphology of the bacteria.

[0056] Removal of impurities: Cell debris, blood components or other impurities in the sample can be removed by means of filtration and centrifugation to improve the purity of bacterial isolation.

[0057] Feature extraction refers to extracting a bacterial growth inhibition feature data group and a drug resistance performance fluctuation feature data group, and then generating a bacterial growth inhibition index based on the bacterial growth inhibition feature data group, and generating a drug resistance performance fluctuation index based on the drug resistance performance fluctuation feature data group.

[0058] Bacterial Growth Inhibition Index: This index is calculated through the growth inhibition characteristic data of bacteria at different drug concentrations and nonlinear regression model to measure the degree of inhibition of bacteria under the action of drugs. This index reflects the sensitivity of bacteria to drugs. The higher the value, the stronger the inhibitory effect of bacteria on drugs, which may mean that the bacteria are sensitive to the drug.

[0059] Resistance Fluctuation Index: This index evaluates the degree of fluctuation in bacterial response to changes in drug concentration by analyzing the fluctuation in bacterial growth performance at different drug concentrations. A higher fluctuation index may indicate that the bacteria are more resistant to the drug because it shows greater adaptability or tolerance to the drug.

[0060] When the test object, i.e., a single strain sample, is automatically distributed to reaction wells containing different drug concentrations, the drug concentrations in adjacent reaction wells change in an isogradient manner and are distributed at the same time.

[0061] Standardization of drug concentration gradient test: Setting the drug concentration to an equal gradient ensures the standardization of the experiment, making the change between different concentrations linear, which helps to compare the sensitivity of different strains to drugs. At the same time, this gradient distribution can avoid errors caused by uneven concentration changes.

[0062] By distributing samples to multiple reaction wells at the same time point, time delays and sample contamination risks during operation are reduced, ensuring consistent conditions in all reaction wells, which is conducive to improving data repeatability and reliability. This gradient distribution method helps to comprehensively evaluate bacterial resistance to drugs at different concentrations. When bacterial growth is completely inhibited, its resistance level can be clearly confirmed, providing a reference for clinical treatment.

[0063] The logic for obtaining the bacterial growth inhibition index is:

[0064] The drug concentration of the i-th reaction well is obtained as C from the bacterial growth inhibition characteristic data set. i , which indicates the concentration of the drug in a certain reaction well, affecting the degree of inhibition of bacterial growth. The bacterial growth performance of the i-th reaction well is divided into G i The score range is [0,1], reflecting the bacterial concentration C i Growth under G i A value approaching 0 indicates that the bacteria are completely inhibited. i The tendency to 1 indicates that the bacteria have fully grown, and then the drug concentration weight W of the i-th reaction well is calculated. i , n represents the total number of reaction wells, p represents the drug concentration adjustment coefficient; drug concentration weight W i The specific degree to which the drug concentration in each reaction well is weighed against the contribution of inhibition is controlled by the concentration adjustment factor p.

[0065] At each drug concentration, the contribution to the inhibition of bacterial growth is introduced with a nonlinear adjustment coefficient λ:

[0066] I i =W i ·(1-G i ) λ ;I i represents the inhibition contribution value of the i-th reaction well, that is, a certain drug concentration C i Contribution to the overall inhibition effect, the nonlinear adjustment coefficient λ controls the nonlinear effect of the bacterial growth inhibition contribution value and is used to adjust the 1-G i sensitivity.

[0067] Fit a nonlinear regression curve through the concentration-growth performance data:

[0068] k is the slope parameter of the fitting curve, indicating that the curve is i The speed of change near the curve, b is the offset parameter of the fitting curve, which determines the position of the curve on the horizontal axis, G(C i ) indicates that the drug concentration is C i When the fitting curve is fitted, the fitting score value is obtained; a nonlinear regression model is established to predict the growth inhibition trend of bacteria.

[0069] Calculate the actual bacterial growth performance score G i and the fitting score G(C i ) to obtain the mean square error E, which is used to measure the accuracy of the fitting model. The smaller the error, the more accurate the model. The formula for calculating the mean square error is not repeated here.

[0070] The formula for calculating the bacterial growth inhibition index is:

[0071] n represents the total number of reflecting holes, E max is the preset maximum error value, which is used to normalize the mean square error E. growth It represents the bacterial growth inhibition index, which is the overall inhibition index after comprehensively considering the inhibition contribution value of all reactions and the fitting error. It indicates the overall inhibitory effect of the drug on bacteria. The higher the index, the stronger the inhibitory effect of the drug on bacteria.

[0072] The drug concentration adjustment coefficient p is obtained based on the α value in pharmacokinetics (PK). The α value usually refers to the rate constant of drug distribution in the body compared to excretion, and is often used in a two-phase model (such as a two-compartment model). One of the common methods of pharmacokinetic models is the two-compartment model, which involves two rate constants: α (distribution phase rate constant) and β (elimination phase rate constant). Specifically, the drug concentration adjustment coefficient p is obtained by the following steps:

[0073] Construct a two-compartment model based on the changes in drug concentration over time in the body:

[0074] C(t)=A·e -αt +B·e -βt ; C(t) represents the concentration of the drug at time t, A and B are constants related to the initial concentration and distribution of the drug, α is the distribution phase rate constant, and β is the elimination phase rate constant. The two-compartment model equation is fitted using the least squares method to solve the values ​​of α and β, which are then substituted into the calculation formula of the drug concentration adjustment coefficient p: The drug concentration adjustment coefficient p reflects the ratio of the distribution phase (α) to the overall rate (α+β), and is used to weigh the relative importance of drug concentration distribution and elimination in the body. The closer p is to 1, the faster the drug distribution rate is, and its inhibitory effect on bacteria is mainly concentrated in the initial high concentration stage. The closer p is to 0, the faster the drug elimination rate is, and its inhibitory effect on bacteria depends more on the long-term continuous effect of low concentrations. The drug concentration weight is calculated based on the concentration adjustment coefficient p. The introduction of the drug concentration adjustment coefficient p makes the weight distribution of high-concentration drugs and low-concentration drugs more reasonable, and can accurately reflect the actual contribution of drugs to the bacterial inhibition effect at different concentrations.

[0075] The logic for obtaining the drug resistance fluctuation index is:

[0076] Obtain the absolute value of the growth performance difference ΔG between two adjacent drug concentrations from the drug resistance performance fluctuation characteristic data set i , which indicates the degree of local fluctuation, is used to quantify the degree of fluctuation between bacterial growth performance in adjacent reaction wells, reflecting the magnitude of change in local bacterial growth. If the fluctuation is large, it means that bacterial growth is sensitive to changes in drug concentration; if the fluctuation is small, it means that bacterial growth is stable.

[0077] Based on the absolute value of growth performance difference ΔG i time series, calculate the overall volatility standard deviation σ of the time series G , the overall volatility standard deviation σ G It is an indicator of the degree of dispersion of the overall fluctuation, which is used to reflect whether the effect of the drug on bacterial growth is consistent. If the standard deviation σ G If the standard deviation σ is large, it means that the fluctuation range is unstable; G Smaller, indicating that the fluctuation range is more stable.

[0078] And the average fluctuation range of the time series Indicates the average level of fluctuation in overall growth performance, used to normalize fluctuation characteristics. The higher the average fluctuation, the greater the fluctuation in bacterial growth performance and the greater the sensitivity to changes in drug concentration.

[0079] Considering the impact of drug concentration changes on the degree of fluctuation, the weight between adjacent drug concentrations in the i-th group is defined as W conc，i :

[0080] |C i+1 -C i | represents the absolute difference between adjacent concentrations, indicating the concentration gradient change, j and i are the indexes of the reaction wells, and n represents the total number of reaction wells; the weight W between adjacent drug concentrations in the i-th group conc，i Reflects the relative impact of changes in drug concentrations in adjacent reaction wells. The larger the change in drug concentration (i.e., |Ci+1 -C i | is larger), the higher the weight is, indicating that the contribution of this concentration change to the fluctuation is more important. By normalizing (denominator), we ensure that the sum of all weights is 1.

[0081] The calculation formula of drug resistance fluctuation index is:

[0082] I var It represents the fluctuation index of drug resistance, which comprehensively considers the influence of local fluctuation, overall fluctuation and drug concentration gradient, and measures the fluctuation degree of bacterial resistance. var When it is higher, it means that the bacterial resistance to drug concentration fluctuates greatly and has poor stability; when I var When it is low, it means that the resistance performance fluctuates less and is more stable.

[0083] The pre-trained machine learning model is a convolutional neural network model, and the output result of the convolutional neural network model is the resistance level of a single strain sample.

[0084] When the convolutional neural network is applied to the drug resistance detection of a single strain sample, the model input includes the characteristic analysis results related to drug resistance, such as the bacterial growth inhibition index and the drug resistance performance fluctuation index, and finally outputs the bacterial resistance level. The following is a detailed explanation of its working principle:

[0085] Data input: The input of the convolutional neural network is preprocessed feature data, including:

[0086] Bacterial Growth Inhibition Index: Measures the overall inhibitory effect of a drug on bacteria.

[0087] Resistance fluctuation index: measures the degree of fluctuation of bacterial resistance at different drug concentrations.

[0088] These feature data are converted into a format suitable for input into a convolutional neural network (such as a two-dimensional matrix, time series, or feature vector).

[0089] The core function of the convolution layer is to extract local features. The convolution operation extracts local patterns of the input data, such as exponential changes, through a sliding filter (or convolution kernel). Each convolution kernel can capture specific patterns or features, such as the relationship between the fluctuation amplitude of drug resistance and the inhibition index. Automatically learn important local features and reduce dependence on manual feature selection. It has translation invariance, that is, local patterns in the data can be captured when they appear in different positions. Subsequently, it passes through the pooling layer and the fully connected layer, and the output of each layer is processed by an activation function (such as ReLU, Sigmoid, etc.) to increase the model's ability to fit complex features. For example, the ReLU function can effectively handle the nonlinear relationship between bacterial growth inhibition and drug resistance fluctuation characteristics.

[0090] The output layer outputs the drug resistance level. The number of neurons in the output layer is equal to the number of drug resistance level categories, such as sensitive (S), intermediate (I), and resistant (R). The output of each neuron represents the probability value of the corresponding drug resistance level. The probability value is normalized to [0,1] through the Softmax activation function. The final output result is the drug resistance level with the largest probability value.

[0091] Embodiment 2: A bacterial resistance detection system, comprising:

[0092] A separation module collects bacterial infection samples and performs preliminary processing on the samples to separate the existing bacteria and obtain processed samples;

[0093] The purification module inoculates the processed sample into a pre-configured culture medium and cultures it under preset environmental conditions. After obtaining a single bacterial colony, it performs purification culture to ensure that the test object is a single strain.

[0094] The detection and recording module uses an automated device to automatically distribute the detection object, i.e., a single strain sample, to reaction wells containing different drug concentrations, and records the performance data of the single strain sample in the reaction wells with different drug concentrations to obtain a performance data set;

[0095] The drug resistance assessment module extracts features based on the performance data set, then performs feature analysis, inputs the feature analysis results into a pre-trained machine learning model, and outputs the drug resistance level of a single strain sample.

[0096] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0097] It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0098] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0099] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0100] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A method for detecting bacterial resistance, characterized in that: The following steps are involved: Collect bacterial infection samples and perform preliminary processing on the samples to isolate the existing bacteria and obtain processed samples; The processed sample is inoculated onto a pre-configured culture medium and cultured under preset environmental conditions. After obtaining a single bacterial colony, purification culture is performed to ensure that the test object is a single strain; Using automated equipment to automatically distribute the test object, i.e., a single strain sample, to reaction wells containing different drug concentrations, and recording the performance data of the single strain sample in the reaction wells with different drug concentrations to obtain a performance data set; Feature extraction is performed based on the performance data set, followed by feature analysis. The feature analysis results are input into a pre-trained machine learning model to output the resistance level of a single strain sample.

2. A method for detecting bacterial resistance according to claim 1, characterized in that: Initial processing includes: centrifugation, filtration, smearing, and removal of impurities.

3. A method for detecting bacterial resistance according to claim 2, characterized in that: Feature extraction refers to extracting a bacterial growth inhibition feature data group and a drug resistance performance fluctuation feature data group, and then generating a bacterial growth inhibition index based on the bacterial growth inhibition feature data group, and generating a drug resistance performance fluctuation index based on the drug resistance performance fluctuation feature data group.

4. A method for detecting bacterial resistance according to claim 3, characterized in that: When the test object, i.e., a single strain sample, is automatically distributed to reaction wells containing different drug concentrations, the drug concentrations in adjacent reaction wells change in an isogradient manner and are distributed at the same time.

5. A method for detecting bacterial drug resistance according to claim 4, characterized in that: The logic for obtaining the bacterial growth inhibition index is: The drug concentration of the i-th reaction well is obtained as C from the bacterial growth inhibition characteristic data set. i , the bacterial growth performance of the i-th reaction well is divided into G i , the score range is [0,1], and then calculate the drug concentration weight W of the i-th reaction well i , n represents the total number of reaction wells, and p represents the drug concentration adjustment coefficient; At each drug concentration, the contribution to the inhibition of bacterial growth is introduced with a nonlinear adjustment coefficient λ: I i =W i ·(1-G i ) λ ; I i represents the inhibition contribution value of the i-th reaction well; Fit a nonlinear regression curve through the concentration-growth performance data: k is the slope parameter of the fitting curve, b is the offset parameter of the fitting curve, G(C i ) indicates that the drug concentration is C i When , the fitting score value obtained by fitting the fitting curve; Calculate the actual bacterial growth performance score G i and the fitting score G(C i ) to obtain the mean square error E; The formula for calculating the bacterial growth inhibition index is: n represents the total number of reflecting holes, E max is the preset maximum error value, which is used to normalize the mean square error E. growth It represents the bacterial growth inhibition index.

6. A method for detecting bacterial resistance according to claim 5, characterized in that: The drug concentration adjustment factor p is obtained by the following steps: Construct a two-compartment model based on the changes in drug concentration over time in the body: C(t)=A·e -αt +B·e -βt ; C(t) represents the concentration of the drug at time t, A and B are constants related to the initial concentration and distribution of the drug, α is the distribution phase rate constant, and β is the elimination phase rate constant. The two-compartment model equation is fitted using the least squares method to solve the values ​​of α and β, which are then substituted into the calculation formula of the drug concentration adjustment coefficient p:

7. A method for detecting bacterial drug resistance according to claim 6, characterized in that: The logic for obtaining the drug resistance fluctuation index is: Obtain the absolute value of the growth performance difference ΔG between two adjacent drug concentrations from the drug resistance performance fluctuation characteristic data set i , indicating the degree of local fluctuation, based on the absolute value of the difference in growth performance ΔG i time series, calculate the overall volatility standard deviation σ of the time series G , and the average fluctuation amplitude of the time series Considering the impact of drug concentration changes on the degree of fluctuation, the weight between adjacent drug concentrations in the i-th group is defined as W conc，i : |C i+1 -C i | represents the absolute difference between adjacent concentrations, indicating the concentration gradient change, j and i are the indexes of the reaction wells, and n represents the total number of reaction wells; The calculation formula of drug resistance fluctuation index is: I var It represents the fluctuation index of drug resistance.

8. A method for detecting bacterial drug resistance according to claim 7, characterized in that: The pre-trained machine learning model is a convolutional neural network model, and the output result of the convolutional neural network model is the resistance level of a single strain sample.

9. A bacterial resistance detection system, based on a bacterial resistance detection method according to any one of claims 1 to 8, characterized in that: include: A separation module collects bacterial infection samples and performs preliminary processing on the samples to separate the existing bacteria and obtain processed samples; The purification module inoculates the processed sample into a pre-configured culture medium and cultures it under preset environmental conditions. After obtaining a single bacterial colony, it performs purification culture to ensure that the test object is a single strain. The detection and recording module uses an automated device to automatically distribute the detection object, i.e., a single strain sample, to reaction wells containing different drug concentrations, and records the performance data of the single strain sample in the reaction wells with different drug concentrations to obtain a performance data set; The drug resistance assessment module extracts features based on the performance data set, then performs feature analysis, inputs the feature analysis results into a pre-trained machine learning model, and outputs the drug resistance level of a single strain sample.

Images