Multi-element antibiotic rapid detection method based on high-throughput array spectrum and deep learning

Abstract

The invention discloses a multi-element antibiotic rapid detection method based on high-throughput array spectrum and deep learning, and belongs to the technical field of water environment new pollutant detection. The method comprises the following steps of antibiotic spectrum sensing system building, combined chemical probe screening, high-throughput spectrum experiment, deep learning stoichiometric model development and antibiotic intelligent detection system development. According to the invention, a high-throughput combined array spectrum sensing technology is adopted, holographic spectrum imaging is taken as a core to obtain fingerprint image data, the experiment efficiency is greatly improved, a high-throughput experiment and a machine learning model are combined, the sampling range is expanded, more types of chemical mixture combinations can be covered, and the detection accuracy is improved. The deep learning model is good at extracting complex features from a large amount of data, fingerprint spectrums of different types of antibiotics can be accurately identified and distinguished, high sensitivity and high specificity can be kept even in a complex water environment, and the accuracy and response speed of pollutant detection in a complex chemical system are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of new pollutant detection in water environment, and specifically to a method for rapid detection of multiple antibiotics based on high-throughput array spectroscopy and deep learning. Background Art

[0002] Antibiotics, due to their potential hazards such as inducing biological resistance genes, destroying the ecological balance of bacterial flora, polluting water sources and threatening human health, have become one of the four typical new pollutants in the current global environmental pollution prevention and control, along with endocrine disruptors, persistent organic pollutants and microplastics. Qualitative and quantitative information of antibiotics in environmental samples such as rivers, soils and groundwater is the basic data source for mastering the spatiotemporal distribution characteristics, migration and transformation laws and ecological risk thresholds of antibiotic pollutants. Existing environmental monitoring technologies have a mature conventional indicator detection system, but due to the accompaniment of pollution behavior and the complexity of environmental media, the traditional "separation first, analysis later" refined detection route faces huge challenges when facing complex antibiotic contaminated samples with a wide range of sources, complex composition and low concentration levels. In recent years, instrumental analysis techniques such as high performance liquid chromatography, fluorescence spectroscopy, capillary electrophoresis, and electrochemiluminescence immunoassay can perform multi-index detection on antibiotic contaminated water samples. However, due to the low concentration of antibiotics in natural water bodies and the complex water matrix, traditional detection technologies face challenges such as complex pre-treatment, cumbersome operation and high cost, which limits their application in actual water body detection. Therefore, the development of intelligent analytical technologies with high sensitivity, high selectivity, high cost-effectiveness, and the ability to simultaneously detect multiple antibiotics is the main trend in the detection of antibiotics in environmental samples in the future.

[0003] At present, the common antibiotic detection methods still have the following problems: (1) Antibiotic pollution in natural water bodies usually comes from a wide range of sources, has a complex composition, and has a low concentration. Traditional antibiotic detection technology often relies on large-scale laboratory instrument detection and analysis. However, the sample pretreatment process of these instruments is complex, the test cycle is long, and the environmental conditions are high. At the same time, the detection equipment is expensive, the operation is cumbersome, and the technical requirements are high. At the same time, the detection process requires a lot of manpower and material resources for sampling, analysis and detection; (2) In the existing AI analysis methods, the process of collecting chemical information of the samples to be tested by instruments and equipment is relatively cumbersome and expensive, and it is difficult to obtain the big data required for training AI models in a short time; (3) When a smaller data set is used for model training, the resulting chemometric model is prone to overfitting, and its robustness and generalization ability are poor, making it difficult to meet the needs of actual application scenarios. It can be seen that when facing antibiotic-contaminated water samples with multiple sources, complex composition, and trace amounts, how to develop intelligent analysis technologies with high sensitivity, high selectivity, and simultaneous detection of multiple antibiotics from the perspectives of hardware sensing and software analysis is the main trend of antibiotic detection in environmental samples in the future.

[0004] Based on this, the present invention combines high-throughput array combined spectroscopy technology with a deep learning algorithm to develop a rapid detection method for multiple antibiotics based on high-throughput array spectroscopy and deep learning. Summary of the invention

[0005] The purpose of the present invention is to provide a method for rapid detection of multiple antibiotics based on high-throughput array spectroscopy and deep learning. A significant advantage of the detection technology based on array spectroscopy deep learning is that it simplifies the sample pre-processing steps, while avoiding the use of a large number of consumables in the data acquisition stage, reducing secondary pollution, and ensuring the accuracy and repeatability of the detection. At the same time, it also significantly reduces the complexity of experimental operations, thereby reducing related detection costs and maintenance costs, and solves the problems of traditional antibiotic detection technology requiring complex pre-processing processes, low data generation, high costs, and difficulty in rapid multi-indicator simultaneous detection.

[0006] The present invention is achieved through the following technical solutions:

[0007] The present invention is a method for rapid detection of multiple antibiotics based on high-throughput array spectroscopy and deep learning, comprising the following steps:

[0008] S1: Construction of antibiotic spectral sensing system

[0009] Among them, the antibiotic spectral sensing system consists of a reaction system and a data collection system. The reaction system includes a holographic spectrometer, and the sample material information is perceived through the holographic spectrometer.

[0010] The data collection system is composed of a highly sensitive computer connected to the reaction system. The digital spectrum images collected by the reaction system are collected and recorded through the collection system.

[0011] S2: Combinatorial chemical probe screening

[0012] By consulting the literature, we selected chemical probes that may react with antibiotics by color difference, divided them into two categories: colorimetric probes and fluorescent probes, and prepared them and antibiotics into standard solutions.

[0013] S3: High-throughput spectroscopy experiments

[0014] The complex chemical solution containing mixed multi-antibiotics was taken as the research object. The chemical information in the substance was enhanced by labeling with multi-component mixed chemical probes and captured in the form of array spectrum by CCD. Then, high-throughput spectral experiments were carried out, and the experimental results were organized, stored and calculated.

[0015] S4: Deep Learning Chemometric Model Development

[0016] Model development includes:

[0017] Dataset standardization,The data set includes image sets and label values. After image acquisition, the images are uniformly standardized.

[0018] The CNN network model was built. Based on the collected mixed antibiotic array spectral image set and its corresponding concentration label set, the SqueezeNet CNN model was used to perform systematic supervised training on the standardized array spectral image set and the corresponding label set.

[0019] CNN model evaluation,For the rapid prediction model of antibiotic concentration, a linear fitting analysis was performed between the antibiotic concentration predicted by the model in the test set and the actual concentration, and the evaluation was performed specifically using the following formula.

[0020]

[0021] Where n is the sample size, is the predicted value, y i is the true value,

[0022] S5: Development of intelligent antibiotic detection system

[0023] Based on the trained CNNModel, we use the Qt Designer tool to create an efficient and intelligent antibiotic rapid detection platform to achieve fast and direct end-to-end prediction of the types and concentrations of antibiotics in water samples.

[0024] Furthermore, the solution preparation process in step S2 includes:

[0025] S2.1 Add different chemical probes and antibiotic solutions

[0026] The addition ratio of chemical probe and antibiotic solution is 1:1.

[0027] S2.2 ELISA Scanner for Colorimetry

[0028] Among them, the absorbance of the colorimetric probe was scanned at 360-780nm, and the fluorescence intensity of the fluorescent probe was scanned at 230-490nm.

[0029] S2.3 Screening of single probes

[0030] Among them, single probes with high sensitivity and selectivity were screened.

[0031] S2.4 Preparation of combinatorial chemical probes

[0032] Among them, single probes with high sensitivity and selectivity are mixed in a certain proportion to prepare a combined chemical probe.

[0033] Furthermore, in step S2.1, the carrier used for dropping different chemical probes and antibiotic solutions is a 96-well plate.

[0034] Furthermore, when different chemical probes and antibiotic solutions were added dropwise, a blank control group without antibiotics was set up.

[0035] Furthermore, in step S3, a series of characteristic fingerprint spectral image sets of multivariate mixed antibiotics are collected in batches by random sampling.

[0036] Furthermore, in step S4, the standardization processing method includes:

[0037] S4.1 rotates, crops, and subtracts blank chemical probes and background noise from the sorted spectral images.

[0038] S4.2 uses the experimental image and the background blank image to make a difference and save the difference image.

[0039] S4.3 Check and process irrelevant data, duplicate data, empty data, abnormal data, erroneous data, etc.

[0040] S4.4 associates the image set with the label set one by one.

[0041] Furthermore, in step S4, when the model is trained, the fingerprint spectrum image containing mixed antibiotics of different concentrations is used as input, and the concentrations of different antibiotics are used as output indicators.

[0042] The present invention has the following beneficial effects:

[0043] First, the high-throughput combined array spectral sensing technology adopted in the present invention obtains fingerprint image data with holographic spectral imaging as the core, which greatly improves the experimental efficiency. This method breaks through the shortcomings of the traditional method of cumbersome experimental process and limited reaction dimension, realizes flexible experiments without input dimension restrictions, significantly expands the scale and diversity of data, provides rich data support for the training of machine learning models, and overcomes the problem of data limitations in traditional methods. By combining high-throughput experiments with machine learning models, the sampling range is effectively expanded to cover more types of chemical mixture combinations. The deep learning model is good at extracting complex features from large amounts of data, and can accurately identify and distinguish the fingerprint spectra of different types of antibiotics. It can maintain high sensitivity and high specificity even in complex water environments, and improve the accuracy and response speed of pollutant detection in complex chemical systems.

[0044] At the same time, the combination of high-throughput experiments and machine learning models can significantly reduce reagent consumption, manpower input and equipment operating time, thereby reducing detection costs. Automated and intelligent technical means enable researchers to obtain and analyze data with higher efficiency, reduce errors caused by human intervention, and improve the repeatability and reliability of experiments. Overall, the prediction results based on deep learning models can assist decision makers in formulating more scientific and reasonable environmental protection strategies and improve the intelligence of environmental monitoring and governance.

[0045] Of course, any product implementing the present invention does not necessarily need to achieve all of the advantages described above at the same time. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 It is a flow chart of the present invention.

[0047] Figure 2 Schematic diagram of the detection system flow.

[0048] Figure 3 Schematic diagram of the high-throughput array spectral sensing system.

[0049] Figure 4 The following are the scanning results of several probes using an ELISA reader.

[0050] Figure 5 This is the prediction result diagram of the SqueezeNet V1.1 model.

[0051] Figure 6 This is the user interface diagram of the antibiotic intelligent detection system platform. DETAILED DESCRIPTION

[0052] 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.

[0053] See also Figure 1-4 The present invention provides a technical solution: a method for rapid detection of multiple antibiotics based on high-throughput array spectroscopy and deep learning, comprising the following steps:

[0054] S1: Construction of antibiotic spectral sensing system

[0055] Among them, the antibiotic spectral sensing system consists of a reaction system and a data collection system. The reaction system includes a holographic spectrometer, and the sample material information is perceived through the holographic spectrometer.

[0056] The holographic spectrometer is composed of a light source, a homogenizer, a filter, a sample reaction pool and a photosensitive coupling element.

[0057] In this embodiment, when the holographic spectrometer is working, when the light emitted by the light source is projected onto the special filter, for each point on the color pattern filter, only light of a specific wavelength can pass through, while light of other wavelengths will be absorbed at this point and cannot pass through. In this way, a specific fingerprint spectrum will be formed after the light source passes through the special filter.

[0058] In this embodiment, the sample reaction pool is where the sample to be tested and the combined chemical probe undergo a color difference reaction. When the spectrum passes through the reaction pool, a specific fingerprint spectrum of the sample pool solution is formed. According to the change in the optical signal caused by the change in the type or concentration of the sample solution in the reaction pool, the array spectrum received by the CCD will also change accordingly. The spectral images obtained from samples of different types and concentrations correspond to them one by one like fingerprints, that is, a specific sample solution will form a specific fingerprint spectrum phase.

[0059] The data collection system is composed of a highly sensitive computer connected to the reaction system. The digital spectrum images collected by the reaction system are collected and recorded through the collection system.

[0060] S2: Combinatorial chemical probe screening

[0061] Taking levofloxacin hydrochloride, tetracycline hydrochloride and dirithromycin as examples, we consulted the literature and selected 29 different types of single probes that may react specifically with them. According to their optical properties, they were divided into two categories: colorimetric probes and fluorescent probes. They were prepared into standard solutions with antibiotics. The names and types are listed in Table 1:

[0062]

[0063]

[0064]

[0065] Table 1. Single probes and their types

[0066] Different chemical probes and antibiotic solutions were dripped in each color block of a 96-well plate in a volume ratio of 1:1, that is, 29 different chemical probes were dripped from top to bottom, and 3 antibiotics were dripped from left to right for specific color difference reaction. A group without antibiotics was set as a blank control group in each group. After the dripping was completed, two types of indicators, colorimetric probes and fluorescent probes, were put into an enzyme reader to scan the absorbance and fluorescence intensity respectively. Specifically, the absorbance of the colorimetric probe at 360-780nm and the fluorescence intensity of the fluorescent probe at 230-490nm were scanned by an enzyme reader, and absorption curves and fluorescence curves were plotted. According to the standards of sensitive color development, large color difference, stable properties and good specificity, 6 chemical probes, namely, cresol red, methyl red, bromothymol blue, sodium fluorescein, eosin y and rhodamine 6G, were screened out, and configured into a combined chemical probe in a certain proportion.

[0067] in, Figure 4 It can be seen that the three colorimetric probes have absorbance curves and the sodium fluorescein has fluorescence curves. Correspondingly, (a) is the absorbance curve graph, and (b) is the sodium fluorescein fluorescence curve graph.

[0068] S3: High-throughput spectroscopy experiments

[0069] Complex chemical solutions containing mixed multi-antibiotics were taken as the research object. The chemical information in the substance was enhanced by labeling with multi-component mixed chemical probes and captured in the form of array spectra through CCD. Spectral experiments were carried out in the antibiotic spectral sensing system using combined chemical probes. The volume, concentration and method of each addition were ensured to be consistent to reduce experimental errors. A total of 1500 characteristic fingerprint images of ternary mixed antibiotics were collected by random sampling with a concentration range of 0-30 mg / L. A ternary characteristic fingerprint dataset was constructed by calculating the label set, and the image set corresponded one to one with the label set.

[0070] S4: Deep Learning Chemometric Model Development

[0071] Model development includes:

[0072] Dataset standardization,The data set includes image sets and label values. After image acquisition, the images are uniformly standardized.

[0073] Among them, the standardization processing method includes rotating and cropping the sorted spectral images, subtracting blank chemical probes and background noise, and making a difference between the experimental image and the background blank image. The saved difference image is the fingerprint spectral image that only contains the internal concentration changes of the mixed antibiotics. At the same time, the data quality is comprehensively improved by checking and processing irrelevant data, duplicate data, empty data, abnormal data, and erroneous data. Finally, the image set and the label set are matched one by one to ensure that the image set and the label set can meet the requirements of subsequent model training.

[0074] The CNN network model was built. Based on the collected mixed antibiotic array spectral image set and its corresponding concentration label set, the SqueezeNet CNN model was used to perform systematic supervised training on the standardized array spectral image set and the corresponding label set.

[0075] Among them, based on the collected mixed antibiotic array spectral image set and its corresponding concentration label set, the SqueezeNet CNN model is used to conduct systematic supervised training on the standardized array spectral image set and the corresponding label set. During model training, the fingerprint spectral image containing mixed antibiotics of different concentrations is used as input, and the concentration of different antibiotics is used as output indicator. 80% of the data set is used as a training set to train the model, and 20% is used as a test set to test the model's prediction results. The trained deep learning model can accurately predict the concentration levels of multiple antibiotics.

[0076] CNN model evaluation,For the rapid prediction model of antibiotic concentration, a linear fitting analysis was performed between the antibiotic concentration predicted by the model in the test set and the actual concentration, and the evaluation was performed specifically using the following formula.

[0077]

[0078] Where n is the sample size, is the predicted value, y i is the true value,

[0079] In the above implementation, based on the constructed data set, data standardization, cleaning, subtraction and other operations are performed to prepare for subsequent model training, and then they are input into the SqueezeNet V1.1CNN network model for supervised training to analyze and predict antibiotics at different concentrations. For the deep learning algorithm concentration prediction model, three indicators, R2, MAE and RMSE, are used to evaluate the model prediction performance.

[0080] like Figure 5As shown in the figure, the results of linear fitting of the predicted values ​​and true values ​​of the SqueezeNet V1.1 network model on the test set of levofloxacin hydrochloride, tetracycline hydrochloride and dirithromycin ternary mixed antibiotic concentration prediction are shown. It can be seen that the fitting between the output predicted values ​​and the true values ​​is relatively excellent overall. The R2, MAE, and RMSE of levofloxacin hydrochloride are 0.926, 1.52, and 1.18 respectively; those of tetracycline hydrochloride are 0.966, 0.972, and 0.783 respectively; and those of dirithromycin are 0.874, 1.88, and 1.50 respectively. These results show that the model shows high accuracy and reliability when predicting the concentrations of multiple antibiotics at the same time. From the comparison of the prediction results of the three antibiotics, the three indicators of the prediction results of tetracycline hydrochloride are better than those of levofloxacin hydrochloride and dirithromycin, which shows that the model can meet the needs of detecting the concentrations of multiple antibiotics at the same time.

[0081] S5: Development of intelligent antibiotic detection system

[0082] like Figure 6 As shown in the figure, an antibiotic intelligent detection system platform built with Qt Designer is demonstrated, which provides users with a convenient operation process through an intuitive user interface. The user only needs to click "Select Image" to upload the array spectrum image to be predicted, and then click "Save Prediction Results" to select the location to output the prediction results. The system will automatically start the prediction model and accurately predict the antibiotic concentration. The prediction results are intuitively presented on the interface. At the same time, the interface supports adaptive resizing, ensuring that users can experience smooth and consistent operations on different devices. This design not only improves the user's operating convenience, but also optimizes the overall interaction process.

[0083] The preferred embodiments of the present invention disclosed above are only used to help illustrate the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to the specific implementation methods described. Obviously, many modifications and changes can be made according to the content of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the present invention, so that those skilled in the art can understand and use the present invention well. The present invention is limited only by the claims and their full scope and equivalents.

Claims

1. A rapid detection method for multiple antibiotics based on high-throughput array spectroscopy and deep learning, characterized in that: The steps include: S1: Construction of antibiotic spectral sensing system The antibiotic spectral sensing system consists of a reaction system and a data collection system. The reaction system includes a holographic spectrometer, which senses sample material information. The data collection system is composed of a highly sensitive computer connected to the reaction system, and the digital spectrum images collected by the reaction system are collected and recorded by the collection system; S2: Combinatorial chemical probe screening By consulting the literature, we selected chemical probes that may react with antibiotics with color difference, and divided them into two categories: colorimetric probes and fluorescent probes. We then prepared them and antibiotics into standard solutions. S3: High-throughput spectroscopy experiments The complex chemical solution containing mixed multi-antibiotics was used as the research object, and the chemical information in the substance was enhanced by multi-component mixed chemical probe labeling, and captured in the form of array spectrum by CCD. Then, high-throughput spectral experiments were carried out, and the experimental results were sorted, stored and calculated. S4: Deep Learning Chemometric Model Development Model development includes: Dataset standardization: The data set includes image sets and label values. After image acquisition, the images are uniformly standardized. The CNN network model was built. Based on the collected mixed antibiotic array spectral image set and the label set of its corresponding concentration, the SqueezeNet CNN model was used to conduct systematic supervised training on the standardized array spectral image set and the corresponding label set. CNN model evaluation: For the antibiotic concentration rapid prediction model, a linear fitting analysis was performed between the antibiotic concentration predicted by the model in the test set and the actual concentration. The evaluation was performed using the following formula: Where n is the sample size, is the predicted value, y i is the true value, S5: Development of intelligent antibiotic detection system Based on the trained CNNModel, we use the Qt Designer tool to create an efficient and intelligent antibiotic rapid detection platform to achieve fast and direct end-to-end prediction of the types and concentrations of antibiotics in water samples.

2. The method for rapid detection of multiple antibiotics based on high-throughput array spectroscopy and deep learning according to claim 1, characterized in that: The holographic spectrometer in step S1 is composed of a light source, a light homogenizer, a filter, a sample reaction pool and a photosensitive coupling element.

3. The method for rapid detection of multiple antibiotics based on high-throughput array spectroscopy and deep learning according to claim 1, characterized in that: The solution preparation process in step S2 includes: S2.1 Add different chemical probes and antibiotic solutions Among them, the addition ratio of chemical probe and antibiotic solution was 1:1; S2.2 ELISA Scanner for Colorimetry Among them, the absorbance of the colorimetric probe was scanned at 360-780nm, and the fluorescence intensity of the fluorescent probe was scanned at 230-490nm; S2.3 Screening of single probes Among them, screening for single probes with high sensitivity and selectivity; S2.4 Preparation of combinatorial chemical probes Among them, single probes with high sensitivity and selectivity are mixed in a certain proportion to prepare a combined chemical probe.

4. The method for rapid detection of multiple antibiotics based on high-throughput array spectroscopy and deep learning according to claim 3, characterized in that: In step S2.1, the carrier used for dropping different chemical probes and antibiotic solutions is a 96-well plate.

5. The method for rapid detection of multiple antibiotics based on high-throughput array spectroscopy and deep learning according to claim 4, characterized in that: When different chemical probes and antibiotic solutions were added dropwise, a blank control group without antibiotics was set up.

6. The method for rapid detection of multiple antibiotics based on high-throughput array spectroscopy and deep learning according to claim 1, characterized in that: In the step S3, a series of characteristic fingerprint spectral image sets of multivariate mixed antibiotics are collected in batches by random sampling.

7. The method for rapid detection of multiple antibiotics based on high-throughput array spectroscopy and deep learning according to claim 1, characterized in that: In step S4, the standardization processing method includes: S4.1 rotates, crops, and subtracts blank chemical probes and background noise from the sorted spectral image; S4.2 Make a difference between the experimental image and the background blank image and save the difference image; S4.3 Check and process irrelevant data, duplicate data, empty data, abnormal data, erroneous data, etc.; S4.4 associates the image set with the label set one by one.

8. The method for rapid detection of multiple antibiotics based on high-throughput array spectroscopy and deep learning according to claim 7, characterized in that: In step S4, when the model is trained, the fingerprint spectrum image containing mixed antibiotics of different concentrations is used as input, and the concentrations of different antibiotics are used as output indicators.

Images