Method for rapidly detecting and identifying pathogenic bacteria by using machine learning enhanced fluorescence sensor array of functionalized graphene quantum dots

Abstract

The invention relates to a method for rapidly detecting and identifying pathogenic bacteria by using a machine learning enhanced fluorescence sensor array based on functionalized graphene quantum dots, which comprises the following steps of: firstly, synthesizing three types of environment-friendly fluorescence graphene quantum dots by a hydrothermal method; secondly, three fluorescent graphene quantum dots are respectively incubated with five pathogenic bacteria for 5 minutes to generate fluorescence intensity difference; then, constructing a sensor array by utilizing the difference of the fluorescence intensities of the three graphene quantum dots; then, identifying single food-borne pathogenic bacteria and mixed food-borne pathogenic bacteria by using a sensor array based on a machine learning algorithm; finally, whether the tap water contains the pathogenic bacteria or not and the types of the pathogenic bacteria can be judged by utilizing an algorithm for rapidly detecting the bacteria. The fluorescent sensor array prepared by the invention can realize rapid and accurate identification and quantification of pathogenic bacteria, shows excellent performance in the aspects of accuracy, sensitivity and stability, and has the characteristics of low cost and high efficiency.

Description

Technical Field

[0001] The present invention belongs to the technical field of microbial detection, and specifically relates to a method for rapid pathogen detection and identification based on a machine learning enhanced fluorescence sensor array of functionalized graphene quantum dots. Background Art

[0002] Bacteria are ubiquitous in nature, characterized by their large numbers and wide distribution. They can cause various infections, food poisoning, and even serious diseases and complications, posing a major threat to human health. Every year, about 7.7 million people die from bacterial infections, accounting for 13.6% of the total number of deaths, making it the second leading cause of death after ischemic heart disease.

[0003] Accurate bacterial identification is essential for timely detection and prevention of infection, especially for distinguishing different bacterial species in mixed samples. Rapid identification of pathogens and their characteristics is crucial in areas such as disease diagnosis, clinical treatment, food processing, and environmental monitoring. In the medical industry, the ability to accurately identify pathogens, their drug resistance, and other key characteristics in a timely manner is essential for medical professionals to develop targeted and effective treatment strategies. Traditional bacterial identification techniques mainly rely on molecular biology methods such as polymerase chain reaction (PCR) and serological analysis, including enzyme-linked immunosorbent assay (ELISA), immunofluorescence, agglutination tests, and radioimmunoassays. Although these methods are reliable and accurate, they are often expensive, time-consuming, and inefficient for rapid detection.

[0004] The application of biosensors is becoming increasingly important in the field of microbiology and bioassays. However, conventional biosensors have significant limitations in terms of bacterial identification. Sensors based on a "lock and key" approach can usually detect only one type of bacteria, which limits their applicability in the analysis of complex samples. This approach relies on highly specific biosignature elements, such as antibodies or receptors, which only bind to specific markers of specific bacteria.

[0005] To solve this problem, researchers have developed fluorescent sensor arrays. As a new type of bacterial detection method, its high sensitivity, selectivity, rapid response and data acquisition capabilities have attracted increasing attention from researchers. They are particularly useful in situations where changes in the concentration of target substances need to be monitored in real time. Fluorescent sensor arrays can detect multiple bacteria simultaneously using a variety of different biosignature elements. These sensors use fluorescent labels and advanced optical detection technology to quickly and sensitively identify and distinguish various bacteria. The advantage of fluorescent sensor arrays lies in their multi-target detection capabilities, which enable them to produce unique fluorescence response patterns for different bacterial species. This multi-parameter detection method not only improves the accuracy of detection, but also enhances the understanding of sample complexity, which is of great value for large-scale screening and real-time monitoring. Fluorescent sensor arrays capture multiple features of the target simultaneously, forming patterns that can be recognized using pattern recognition and machine learning algorithms.

[0006] Graphene quantum dots (GQDs) are an emerging class of carbon-based fluorescent materials with unique physical and chemical properties that have attracted great attention in recent years in scientific research and practical applications. In general, GQDs are below 100 nanometers in size and several nanometers in thickness, and consist of one, two, or several layers of graphene structures, which are essentially very small graphene sheets. They are structurally stable, resistant to strong acids and bases, and can maintain their performance even in harsh environments. Notably, GQDs do not contain toxic metal elements such as cadmium and lead, and are harmless to the environment. GQDs have wide band gaps, high carrier mobility, and strong luminescence properties, and their fluorescence emission spans a wide spectral range, including ultraviolet (UV), visible light, and infrared (IR). Due to their nanoscale size, GQDs exhibit a significant quantum confinement effect, which limits the energy of electrons and holes, resulting in unique optical and electronic properties. The edges of GQDs contain functional groups that significantly affect their properties, and their properties can be effectively tuned by edge modification. GQDs have stable photoluminescence properties, and their fluorescence emission wavelength and intensity can be adjusted by changing their size, shape and surface modification. These properties make GQDs widely used in bioimaging, optoelectronic devices, chemical and biological sensors, and semiconductor materials. In order to improve the luminescence performance and quantum yield of GQDs, heteroatom doping is an effective method. Doping with elements such as sulfur and nitrogen can change the electron density and reactivity of GQDs, usually leading to improved fluorescence properties. Sulfur-nitrogen co-doped GQDs especially show excellent fluorescence properties.

[0007] Machine learning (ML) is a branch of artificial intelligence (AI) that aims to build systems that can learn or improve performance based on the data they are fed. Machine learning algorithms are based on data, enabling computers to learn from data and make predictions or decisions without being explicitly programmed. This field covers a wide range of applications from the study of statistical and computational techniques to simulating the way humans learn, covering almost all industries and fields, such as healthcare, finance, image recognition and processing, etc. Machine learning can improve the accuracy of decision-making, optimize processes, reduce costs, and in some cases achieve automation, thereby promoting innovation and development in all walks of life. As the amount of data grows and computing power increases, machine learning will continue to play an increasingly important role in scientific research, business applications, and social life.

[0008] The application of machine learning algorithms in the field of bacterial detection has shown significant advantages over traditional detection methods, including improved efficiency and automation, improved decision-making, the ability to process large amounts of data, and the potential for cost savings. In addition, the machine algorithm can also accurately classify and identify foodborne pathogenic bacteria, providing a more comprehensive and accurate guarantee for food safety. Summary of the invention

[0009] The object of the present invention is to provide a method for rapid pathogen detection and identification using a machine learning enhanced fluorescence sensor array of functionalized graphene quantum dots.

[0010] To achieve the above-mentioned purpose and other related purposes, the technical solution provided by the present invention is: a method for rapid pathogen detection and identification based on a machine learning enhanced fluorescence sensor array of functionalized graphene quantum dots, comprising the following steps: S1: synthesizing antibiotic-modified graphene quantum dots, measuring the fluorescence emission spectrum of the antibiotic-modified graphene quantum dots, selecting excitation and emission wavelengths, and measuring the fluorescence intensity of the antibiotic-modified graphene quantum dots at the selected wavelengths; S2: A single colony of the pathogen growing on an agar plate containing LB medium was washed with phosphate buffered saline, resuspended and measured with a spectrophotometer at 600 nm. 600 For the standard curve, the pathogen suspension was diluted to different concentrations with PBS buffer; S3: construct a fluorescent sensor array to combine pathogen suspension with antibiotic-modified graphene quantum dots; S4: The obtained fluorescence response data were normalized and analyzed based on machine learning methods; each graphene quantum dot of the pathogen was systematically replicated multiple times to generate a comprehensive training data set; the classification algorithm built into MATLAB was trained using a random selection procedure to assign 70% of the available data as a training set and the remaining 30% as a test set, without assuming that one algorithm was superior to the others; the five algorithms, k-nearest neighbor algorithm, naive Bayes, decision tree, linear discriminant analysis, and support vector machine, were able to successfully separate the bacteria.

[0011] The preferred technical solution is: in step S1, the antibiotic-modified graphene quantum dots include spectinomycin-modified graphene quantum dots, kanamycin-modified graphene quantum dots and polymyxin B-modified graphene quantum dots.

[0012] The preferred technical solution is: in step S1, the method for synthesizing antibiotic-modified graphene quantum dots comprises: (1) Synthesis of spectinomycin-modified graphene quantum dots Anhydrous citric acid, thiourea and spectinomycin hydrochloride were added to water, stirred, and then the mixture was transferred to a Teflon-lined autoclave and heated at 140-170°C for 1-10 hours; (2) Synthesis of kanamycin-modified graphene quantum dots Anhydrous citric acid, L-cysteine ​​and kanamycin sulfate were added to water, stirred, and then heated in a Teflon-lined reactor at 150-200° C. for 5-15 hours; (3) Synthesis of polymyxin B-modified graphene quantum dots Anhydrous citric acid, sulfur powder, polymyxin B and ammonia water were added to water, stirred, and then heated at 150-200°C in a Teflon-lined autoclave for 10-18 hours.

[0013] The preferred technical solution is: measuring the fluorescence intensity of graphene quantum dots modified with spectinomycin, graphene quantum dots modified with kanamycin and graphene quantum dots modified with polymyxin B under the conditions of excitation wavelength of 300-440 nm and emission wavelength of 300-700 nm.

[0014] The preferred technical scheme is: determining the graphene quantum dots modified with spectinomycin, the graphene quantum dots modified with kanamycin and the graphene quantum dots modified with polymyxin B, and determining the presence of C, N, O and S components in the graphene quantum dots modified with spectinomycin, the graphene quantum dots modified with kanamycin and the graphene quantum dots modified with polymyxin B by X-ray photoelectron spectroscopy, as well as the Zeta potential of the three materials.

[0015] 6. The method for rapid pathogen detection and identification based on a machine learning enhanced fluorescence sensor array of functionalized graphene quantum dots according to claim 1, characterized in that: in step S3, the binding time is 3-8 min.

[0016] The preferred technical solution is: in step S4, the fluorescence response data is the change in fluorescence intensity after the spectinomycin-modified graphene quantum dots, kanamycin-modified graphene quantum dots and polymyxin B-modified graphene quantum dots are mixed with pathogenic bacteria, wherein the formula for normalizing the obtained fluorescence response data is: (F0-F) / F0; Where F0 is the fluorescence intensity before adding pathogens, and F is the fluorescence intensity after adding pathogens.

[0017] The preferred technical solution is: pathogenic bacteria include Salmonella typhimurium, Escherichia coli O157:H7 , Staphylococcus aureus, Pseudomonas aeruginosa, and Listeria monocytogenes.

[0018] The preferred technical solution is: a total of thirteen repeated experiments were conducted for each combination of graphene quantum dots and pathogens, thereby constructing a training data set, which is calculated by multiplying three graphene quantum dots by five pathogenic bacteria and then multiplying by thirteen repetitions.

[0019] Due to the application of the above technical solution, the present invention has the following advantages compared with the prior art: 1. The fluorescent sensor array used in the present invention has a simple synthesis method, a green and environmentally friendly synthesis process, does not involve toxic heavy metal elements, has a low platform production cost, and has good economic benefits and market application prospects.

[0020] 2. The method proposed in the present invention can rapidly and accurately detect pathogenic bacteria in food.

[0021] 3. The algorithm used in the present invention is based on easy-to-operate software implementation, and only requires training and analysis using the built-in classification algorithm of MATLAB to effectively distinguish various types of bacteria.

[0022] 4. Compared with the prior art, the present invention exhibits excellent specificity, significantly shortens the detection time, simplifies the operation steps, and reduces the detection cost. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 TEM images of three graphene quantum dots in Example 1; lattice spacing image; Zeta potential image.

[0024] Figure 2 XPS graph of three graphene quantum dots in Example 1; FT-IR graph.

[0025] Figure 3 The fluorescence emission spectra of the three graphene quantum dots in Example 2; the three graphene quantum dots and five single bacteria (Escherichia coli) at different concentrations O157:H7 , Staphylococcus aureus, Salmonella typhimurium, Pseudomonas aeruginosa and Listeria monocytogenes. The concentration range of the five bacteria is 1.0×10 3 CFU / mL - 1.0×10 7 CFU / mL.

[0026] Figure 4 The relative fluorescence intensity changes of the three graphene quantum dots reacting with the five bacteria in Example 3; the recognition efficiency of the five algorithms for a single bacterial sample; the confusion matrix of the three classification algorithm results of the KNN algorithm for a single bacterial sample; and the parallel coordinates of the three typical factors (factor 1, factor 2, factor 3) of a single bacterial sample. The bacteria were all within 1.0 × 10 3 CFU / mL concentration.

[0027] Figure 5 The characteristic space map of the five bacteria drawn by machine learning in Example 3; the standardized score map for identifying the five bacterial samples; the ring cluster heat map of the five bacteria; and the radar map of the five bacteria. 3 CFU / mL concentration.

[0028] Figure 6 For the different concentrations of Escherichia coli in Example 4 O157:H7 Normalized score plot of factor 1 and Escherichia coli O157:H7 Linear relationship diagram of concentration; Escherichia coli at different concentrations O157:H7 Ring cluster heat map of Escherichia coli at different concentrations O157:H7 Radar chart. The bacteria were all at 1.0 × 10 3 CFU / mL -1.0 × 10 7 CFU / mL concentration.

[0029] Figure 7 The standardized score diagram of different concentrations of Staphylococcus aureus in Example 4; the linear relationship diagram between factor 1 and the concentration of Staphylococcus aureus; the heat map of the ring clustering of Staphylococcus aureus at different concentrations; and the radar map of Staphylococcus aureus at different concentrations. 3 CFU / mL -1.0 × 10 7 CFU / mL concentration.

[0030] Figure 8The standardized score diagram of different concentrations of Salmonella typhimurium in Example 4; the linear relationship diagram between factor 1 and the concentration of Salmonella typhimurium; the circular clustering heat map of Salmonella typhimurium at different concentrations; and the radar map of Salmonella typhimurium at different concentrations. 3 CFU / mL -1.0 × 10 7 CFU / mL concentration.

[0031] Fig. 9 The standardized score diagram of different concentrations of Pseudomonas aeruginosa in Example 4; the linear relationship diagram between factor 1 and Pseudomonas aeruginosa concentration; the circular clustering heat map of Pseudomonas aeruginosa at different concentrations; the radar map of Pseudomonas aeruginosa at different concentrations. The bacteria were all at 1.0 × 10 3 CFU / mL -1.0 × 10 7 CFU / mL concentration.

[0032] Fig.10 The standardized score diagram of Listeria monocytogenes at different concentrations in Example 4; the linear relationship diagram between factor 1 and Listeria monocytogenes concentration; the circular clustering heat map of Listeria monocytogenes at different concentrations; and the radar map of Listeria monocytogenes at different concentrations. 3 CFU / mL -1.0 × 10 7 CFU / mL concentration.

[0033] Fig.11 This is a typical scoring diagram for identifying two mixed bacteria, Escherichia coli O157:H7 and Pseudomonas aeruginosa, using the fluorescence array method in Example 5; O157:H7 Typical scoring graph for a mixture of two bacteria: Listeria monocytogenes; Escherichia coli O157:H7 Typical scoring chart of three mixed bacteria: Pseudomonas aeruginosa, Listeria monocytogenes, and Pseudomonas aeruginosa. The above bacterial concentrations are all 1.0 × 10 3 CFU / mL.

[0034] Fig.12 The anti-interference and stability of the fluorescence platform in Example 6. Typical score diagram of the platform distinguishing five bacteria and some interfering substances (Na+, K+, Cl−, NO3−, Glucose, Galacto- ose, Arginine, Lysine); the relationship between the fluorescence signal and the number of repetitions (n=20); the relationship between the fluorescence signal and the storage time (n=15). The bacterial concentration was 1.0 ×10 3 CFU / mL.

[0035] Fig.13In Example 7, the fluorescence array method was used to evaluate the blind samples consisting of single bacteria, binary mixtures, and ternary mixtures of five bacteria in PBS, and the bacterial concentration was 1.0 × 10 3 CFU / m.

[0036] Fig.14 This is a typical score diagram of the five bacterial differentiation platforms in tap water in Example 8. The bacterial concentration is 1.0× 10 3 CFU / m.

[0037] Fig.15 This is the flow chart of the K-nearest neighbor algorithm (KNN).

[0038] Fig.16 This is the flow chart of the Naive Bayes algorithm.

[0039] Fig.17 This is the flow chart of the decision tree algorithm.

[0040] Fig.18 This is the flowchart of the linear discriminant analysis algorithm (LDA).

[0041] Fig.19 This is the flow chart of the support vector machine (SVM) algorithm.

[0042] Fig. 20 It is a flow chart of the present invention. DETAILED DESCRIPTION

[0043] The following is a description of the implementation of the present invention by means of specific embodiments. Those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in the embodiments.

[0044] See also Figure 1-20 . It should be noted that the structures, proportions, sizes, etc. illustrated in the drawings of this specification are only used to match the contents disclosed in the specification for people familiar with this technology to understand and read, and are not used to limit the limiting conditions under which the present invention can be implemented. Therefore, they have no substantive technical significance, and any structural modification, change in proportional relationship or adjustment of size. The following examples are provided to better understand the present invention, but are not intended to limit the present invention. The experimental methods in the following examples are all conventional methods unless otherwise specified. The experimental materials used in the following examples are all purchased from conventional biochemical reagent stores unless otherwise specified.

[0045] Example 1: Preparation and characterization of three graphene quantum dots.

[0046] 1. Preparation of three graphene quantum dots (S-NS-GQDs, K-NS-GQDs, and B-NS-GQDs): When synthesizing S-NS-GQDs, 0.84 g of anhydrous citric acid, 1.24 g of thiourea, and 0.28 g of spectinomycin hydrochloride were dissolved in 20 mL of ultrapure water and stirred under a magnetic stirrer for 5 min. The mixture was then transferred to a Teflon-lined autoclave and heated at 160 °C for 6 h. 4.2 g of anhydrous citric acid, 2 g of L-cysteine, and 1.4 g of kanamycin sulfate were dissolved in 20 mL of ultrapure water, stirred for 5 min, and heated at 180 °C for 10 h in a Teflon-lined reactor to prepare K-NS-GQDs. 0.42 g of anhydrous citric acid, 0.3 g of sulfur powder, 0.14 g of polymyxin B, and 7.5 mL of 28% ammonia water were dissolved in 7.5 mL of ultrapure water, stirred for 5 min, and heated at 180 °C for 12 h in a Teflon-lined autoclave to obtain B-NS-GQDs. After cooling to room temperature, the three synthesized graphene quantum dots were filtered twice through a 0.22 μm polyethersulfone membrane and stored at 4 °C.

[0047] 2. Characterize the three graphene quantum dots: ① Observe their morphology using field emission transmission electron microscopy (TEM). Figure 1 .from Figure 1 It can be seen that the sizes of the three graphene quantum dots are between 4-7 nm, the lattice spacing is between 0.20-0.26 nm, and the three graphene quantum dots are all negatively charged; ② XPS and FT-IR are used to detect their functional groups. The results are as follows Figure 2 .from Figure 2 It can be seen that these X-ray photoelectron spectroscopy (XPS) results show that N and S have been successfully doped into GQDs, and Fourier transform infrared spectroscopy (FT-IR) results show that there are a large number of amino and hydroxyl groups on the surface of NS-GQDs, which makes the three graphene quantum dots (NS-GQDs) have good hydrophilicity. The above characterization results prove that the three graphene quantum dots have been successfully prepared.

[0048] Example 2: Fluorescence responses of three graphene quantum dots to bacteria.

[0049] (1) Experimental method: Using Escherichia coli O157:H7 (ATCC 9637) was the research object. Single colonies of the five bacteria grown on Luria-Bertani (LB) agar plates were washed twice with phosphate-buffered saline (PBS; 3 mL). The resuspended bacteria were then quantified using a spectrophotometer to measure the OD value at 600 nm. According to the pre-established CFU-OD 600 For the standard curve, the bacterial suspension was diluted with PBS to different concentrations (1.0×10 3CFU / mL, 1.0×10 4 CFU / mL, 1.0×10 5 CFU / mL, 1.0×10 6 CFU / mL, 1.0×10 7 CFU / mL) for later use. O157:H7 The suspension was combined with three types of graphene quantum dots for 5 min, and the fluorescence changes were measured by fluorescence spectrophotometer.

[0050] (2) Experimental results: Figure 3 .from Figure 3 As can be seen in the figure, the fluorescence emission spectra of three graphene quantum dots.

[0051] It was found that fluorescence quenching was significantly related to bacterial concentration. The fluorescence intensity of the three graphene quantum dots increased with O157:H7 The fluorescence quenching efficiency of graphene quantum dots decreased with increasing concentration, with the most significant decrease occurring when the bacterial concentration was the highest. This clear link between fluorescence quenching efficiency and bacterial concentration is critical to understanding how graphene quantum dots interact with bacteria, suggesting their potential use as a quantitative detection method.

[0052] Example 3: Performance test of the fluorescence sensor array in distinguishing five types of bacteria.

[0053] (1) Experimental method: Salmonella typhimurium and Escherichia coli were selected O157:H7 Five bacteria, Staphylococcus aureus, Pseudomonas aeruginosa, and Listeria monocytogenes, were selected as the research objects. Single colonies of the five bacteria grown on Luria-Bertani (LB) agar plates were washed twice with phosphate-buffered saline (PBS; 3 mL). The resuspended bacteria were then quantified using a spectrophotometer to measure the OD value at 600 nm. According to the pre-established CFU-OD 600 For the standard curve, five bacterial suspensions were diluted to 1.0 × 10 3 CFU / mL, five different bacterial suspensions were combined with three types of graphene quantum dots for 5 min, and the fluorescence changes were measured by fluorescence spectrophotometer.

[0054] (2) Experimental results: ① The results are as follows Figure 4 .from Figure 4 It can be seen that the three NS-GQDs sensing units showed unique fluorescence response patterns to these bacteria, thus achieving effective differentiation; the five selected machine learning algorithms all showed excellent performance; the TPR and PPV of the confusion matrix both reached 100%, indicating that the accuracy of the machine learning algorithm was high; the parallel coordinates diagram illustrates the relationship between the three key factors identified by the machine learning ML algorithm, which can be visually represented. ② The results are shown in Figure 5 .from Figure 5 It can be seen that the five bacteria can be distinguished in the three-dimensional feature space, and the five different bacteria have their own clusters; based on the two most significant factors, a bacterial score map was constructed, and the score map has a significant distinction effect; the data was processed by clustering method, and the ring cluster heat map was used to describe the different groups of five bacteria, and the distinction effect was obvious; the radar map of the five bacteria highlights the difference in fluorescence signals when the five bacteria are combined with the three graphene quantum dots, providing strong visual support for bacterial identification. In general, the three graphene quantum dots have good performance in distinguishing the five bacteria.

[0055] Example 4: Testing of the discrimination performance of the fluorescent sensor array on single bacteria of different concentrations.

[0056] (1) Experimental method: Select Escherichia coli O157:H7 For the study, single colonies grown on Luria-Bertani (LB) agar plates were washed twice with phosphate-buffered saline (PBS; 3 mL), and the resuspended bacteria were quantified using a spectrophotometer to measure the OD value at 600 nm. 600 For the standard curve, the bacterial suspension was diluted with PBS to different concentrations (1.0×10 3 CFU / mL, 1.0×10 4 CFU / mL, 1.0×10 5 CFU / mL, 1.0×10 6 CFU / mL, 1.0×10 7 CFU / mL) were set aside, and bacterial suspensions of different concentrations were combined with the three types of graphene quantum dots for incubation for 5 min, and then the fluorescence changes were measured using a fluorescence spectrophotometer. The experimental methods for the other four bacteria were the same as above.

[0057] (2) Experimental results: Escherichia coli O157:H7 The results are as follows Figure 6 .from Figure 6 As can be seen from the score diagram, the sensor array can effectively distinguish different concentrations (1.0×10 3 CFU / mL, 1.0×10 4 CFU / mL, 1.0×10 5 CFU / mL, 1.0×10 6 CFU / mL, 1.0×10 7 CFU / mL) of Escherichia coli O157:H7 ; Fluorescence intensity of sensor array and E. coli O157: H7 There is a strong linear relationship between the concentrations; the circular cluster heat map shows the effects of three graphene quantum dots on different concentrations of E. coli O157:H7, the distinction effect is obvious; the radar chart also highlights the different concentrations of E. coli O157:H7 The difference in fluorescence signals when combined with the three graphene quantum dots. In general, the three graphene quantum dots have good discrimination for individual bacteria at different concentrations. The results for Staphylococcus aureus are as follows Figure 7 . Results for Salmonella typhimurium Figure 8 The results for Pseudomonas aeruginosa are as follows Fig. 9 Listeria monocytogenes results Fig.10 The single bacteria differentiation effect is obvious at five different concentrations Example 5: Performance test of fluorescent sensor array in identifying mixed bacteria.

[0058] (1) Experimental method: Prepare bacterial suspensions of two or three different bacteria for testing recognition performance. ① Select Escherichia coli O157:H7, Pseudomonas aeruginosa was used as the research object, and a bacterial suspension of the two bacteria was prepared. Two single colonies grown on Luria-Bertani (LB) agar plates were washed twice with phosphate buffered saline (PBS; 3 mL). O157:H7 aeruginosa at different ratios (1.0 + 0, 0.8 + 0.2, 0.6 + 0.4, 0.5 + 0.5, 0.4 + 0.6, 0.2 + 0.8, 0 + 1.0), and the resuspended bacteria were quantified by spectrophotometer, and the OD value at 600 nm was measured according to the pre-established CFU-OD 600 For the standard curve, dilute the final suspension of the two bacteria to 1.0 × 10 3 CFU / mL, the bacterial suspension was combined with three types of graphene quantum dots and incubated for 5 min, and then the fluorescence change was measured by fluorescence spectrophotometer. ② Select Escherichia coli O157:H7, Listeria monocytogenes was used as the research object, and a bacterial suspension of the two bacteria was prepared. The operation method was the same as ①, and the final suspension of the two bacteria was also diluted to 1.0×10 3 CFU / mL. ③Select Escherichia coli O157:H7, Pseudomonas aeruginosa and Listeria monocytogenes were used as the research objects, and a bacterial suspension of the three bacteria was prepared. The operation method was the same as ①, and the final suspension of the two bacteria was also diluted to 1.0×10 3 CFU / mL.

[0059] (2) Experimental results: Figure 7 .from Figure 7 Escherichia coli O157:H7 The standard scoring graph of the mixture with Pseudomonas aeruginosa shows that the array can clearly identify bacteria in both mixtures; Escherichia coli O157:H7The standard scoring graph of a sample of a mixture of Listeria monocytogenes and Escherichia coli in the same manner also showed that the array could clearly identify bacteria in the mixture; O157:H7, The scoring standard diagram for Pseudomonas aeruginosa and Listeria monocytogenes also has a clear effect in distinguishing between them.

[0060] Example 6: Analysis of anti-interference ability and stability of fluorescence sensor array.

[0061] (1) Experimental method: The anti-interference ability of the platform was studied. The amount of interference solution and bacterial suspension used in the experiment was the same, both 0.5 mL. Interference substances NaOH, KOH, HCl, HNO3, Glucose, Galactose, Arginine, Lysine were selected and prepared into solutions with a concentration of 25 µM; Salmonella typhimurium and Escherichia coli were selected. O157:H7 Five bacteria, Staphylococcus aureus, Pseudomonas aeruginosa, and Listeria monocytogenes, were selected as the research objects. Single colonies of the five bacteria grown on Luria-Bertani (LB) agar plates were washed twice with phosphate-buffered saline (PBS; 3 mL). The resuspended bacteria were then quantified using a spectrophotometer to measure the OD value at 600 nm. According to the pre-established CFU-OD 600 For the standard curve, five bacterial suspensions were diluted to 1.0 × 10 3 CFU / mL. The five different bacterial suspensions and eight interference solutions were combined with the three types of graphene quantum dots for 5 min, and the fluorescence changes were measured by fluorescence spectrophotometer.

[0062] (2) Experimental results: Figure 8 As shown in the figure, the selected interfering substances (Na+, K+, Cl−, NO3−, Glucose, Galactose, Arginine, Lysine) can be completely distinguished from the five bacteria, indicating that the platform has excellent anti-interference performance; and the repeatability and long-term stability of the platform were evaluated. After twenty consecutive measurements, the fluorescence intensity attenuation of the platform can be ignored; even after being stored for fifteen days, the fluorescence signal of the platform remains stable; the results show that the platform has good sensitivity, anti-interference and stability, and has good practical application prospects.

[0063] Example 7: Detection of real samples by fluorescence sensor array.

[0064] (1) Experimental method: The practicality of the platform was evaluated using blind samples consisting of single bacteria, binary mixtures, and ternary mixtures of five bacterial species. The total concentration of all samples was 1.0×10 3CFU / mL. ① The experimental method for single bacterial samples of five kinds of bacteria is as follows. O157:H7 Five bacteria, Staphylococcus aureus, Pseudomonas aeruginosa, and Listeria monocytogenes, were selected as the research objects. Single colonies of the five bacteria grown on Luria-Bertani (LB) agar plates were washed twice with phosphate-buffered saline (PBS; 3 mL). The resuspended bacteria were then quantified using a spectrophotometer to measure the OD value at 600 nm. According to the pre-established CFU-OD 600 For the standard curve, five bacterial suspensions were diluted to 1.0 × 10 3 CFU / mL, five different bacterial suspensions were combined with three types of graphene quantum dots for 5 min, and the fluorescence changes were measured by fluorescence spectrophotometer. ② The experimental method of mixed bacterial samples is as follows. O157:H7, Pseudomonas aeruginosa was used as the research object, and a bacterial suspension of the two bacteria was prepared. Two single colonies grown on Luria-Bertani (LB) agar plates were washed twice with phosphate buffered saline (PBS; 3 mL). O157: H7 and Pseudomonas aeruginosa in different ratios (1.0 + 0, 0.5 + 0.5, 0 + 1.0), and the resuspended bacteria were quantified by spectrophotometer, and the OD value at 600 nm was measured. 600 For the standard curve, dilute the final suspension of the two bacteria to 1.0 × 10 3 CFU / mL, the bacterial suspension was combined with three types of graphene quantum dots and incubated for 5 min, and then the fluorescence change was measured by fluorescence spectrophotometer. ③ Select Escherichia coli O157:H7, Listeria monocytogenes was used as the research object, and a bacterial suspension of the two bacteria was prepared. The operation method was the same as ②, and the final suspension of the two bacteria was diluted to 1.0×10 3 CFU / mL. ④Select Escherichia coli O157:H7, Pseudomonas aeruginosa and Listeria monocytogenes were used as the research objects, and a bacterial suspension of the three bacteria was prepared. The operation method was the same as ②, and the Escherichia coli O157:H7 , Pseudomonas aeruginosa and Listeria monocytogenes were mixed in different proportions (1.0 + 0 + 0, 0.2 + 0.2 + 0.6, 0 + 1.0 + 0), and the final suspension of the two bacteria was diluted to 1.0 × 10 3 CFU / mL.

[0065] (2) Experimental results: Fig. 9As shown, a total of seventy blind samples were accurately identified, including thirty-six single-bacteria samples, twenty-three binary mixed bacteria samples, and eleven ternary mixed bacteria samples.

[0066] Example 8: Application of fluorescent sensor array to detect bacteria in tap water.

[0067] (1) Experimental method: 500 mL of tap water was sterilized and five types of bacteria were inoculated into the tap water for overnight culture. After overnight culture, the bacterial suspension was diluted to 1.0×10 3 CFU / mL was reserved. 0.5 mL of the collection solution of the five bacterial suspensions of the same batch was incubated with 0.5 mL of graphene quantum dots for 5 min, and the fluorescence intensity value was measured with a fluorescence spectrophotometer.

[0068] (2) Experimental results: Fig.10 ;from Fig.10 It can be seen that five types of bacteria can be identified using machine learning, and the score graphs of the five types of bacteria are clearly classified.

[0069] We have successfully developed a graphene quantum dot (GQDs)-based fluorescent sensor array for rapid and accurate detection of pathogenic bacteria in food. By modifying GQDs with different antibiotics, we were able to specifically identify a variety of bacteria. Experimental results show that the sensor array is superior to traditional methods in terms of detection efficiency, accuracy, and cost-effectiveness.

[0070] The machine learning algorithm used in the present invention is based on MATLAB software, and the required hardware conditions only need to meet the 64-bit operating system of Windows 10, 16GB of memory, and a graphics card of Nvidia GTX 1050Ti. Using machine learning algorithms to detect bacteria is an innovative and practical method that can provide more comprehensive and accurate protection for environmental safety.

[0071] The algorithm database of the present invention is obtained by: (1) Mix 0.5 mL of the five bacterial suspensions with 0.5 mL of graphene quantum dots. After incubation for 5 min, measure the fluorescence intensity using a fluorescence spectrophotometer.

[0072] (2) The training matrix was constructed using the (F0-F) / F0 value, where F0 represents the initial fluorescence intensity of the graphene quantum dot solution and F represents the fluorescence intensity after the bacteria and graphene quantum dots were mixed.

[0073] (3) The data generated by the array is further processed by customized machine learning algorithms, including the k-nearest neighbor algorithm (KNN), the naive Bayes algorithm, the decision tree algorithm, the linear discriminant analysis (LDA) algorithm, and the support vector machine (SVM) algorithm; the flowcharts of the relevant algorithms are shown in the figure. Figure 15-19 shown.

[0074] A method for rapidly detecting bacteria in meat based on fluorescent graphene quantum dots and machine learning algorithms is obtained.

[0075] (1) Thirteen replicates were performed for each NS-GQDs on each bacterium to construct a comprehensive training dataset (three NS-GQDs × 5 bacteria × 13 replicates).

[0076] (2) Use the built-in classification algorithm in MATLAB for training. Randomly select 70% of the data as the training set and the remaining 30% as the test set. Do not assume the superiority of any algorithm.

[0077] (3) Five algorithms, namely k-nearest neighbor (KNN), naive Bayes, decision tree, linear discriminant analysis (LDA) and support vector machine (SVM), can successfully distinguish bacteria.

[0078] (4) Among the five algorithms, the k-nearest neighbor algorithm (KNN) calculates the distance between the point to be classified and all other points, which is usually measured using Euclidean distance or Manhattan distance. The KNN algorithm is particularly suitable for classification and regression problems due to its simplicity of principle, unlimited flexibility for data types, and no need to make any assumptions about data distribution. Its wide adaptability to data forms further enhances the flexibility of the KNN algorithm in practical applications.

[0079] The above description is only used to explain the preferred embodiments of the present invention, and is not intended to limit the present invention in any form. Therefore, any modifications or changes made to the present invention under the same inventive spirit should still be included in the scope of protection intended by the present invention.

Claims

1. A method for rapid pathogen detection and identification based on a machine learning enhanced fluorescence sensor array of functionalized graphene quantum dots, characterized in that: The following steps are involved: S1: synthesizing antibiotic-modified graphene quantum dots, measuring the fluorescence emission spectrum of the antibiotic-modified graphene quantum dots, selecting excitation and emission wavelengths, and measuring the fluorescence intensity of the antibiotic-modified graphene quantum dots at the selected wavelengths; S2: A single colony of the pathogen growing on an agar plate containing LB medium was washed with phosphate buffered saline, resuspended and measured with a spectrophotometer at 600 nm. 600 For the standard curve, the pathogen suspension was diluted to different concentrations with PBS buffer; S3: construct a fluorescent sensor array to combine pathogen suspension with antibiotic-modified graphene quantum dots; S4: The obtained fluorescence response data were normalized and analyzed based on machine learning methods; each graphene quantum dot of the pathogen was systematically replicated multiple times to generate a comprehensive training data set; the classification algorithm built into MATLAB was trained using a random selection procedure to assign 70% of the available data as a training set and the remaining 30% as a test set, without assuming that one algorithm was superior to the others; the five algorithms, k-nearest neighbor algorithm, naive Bayes, decision tree, linear discriminant analysis, and support vector machine, were able to successfully separate the bacteria.

2. The method for rapid pathogen detection and identification based on a machine learning enhanced fluorescence sensor array of functionalized graphene quantum dots according to claim 1, characterized in that: In step S1, the antibiotic-modified graphene quantum dots include spectinomycin-modified graphene quantum dots, kanamycin-modified graphene quantum dots and polymyxin B-modified graphene quantum dots.

3. The method for rapid pathogen detection and identification based on a machine learning enhanced fluorescence sensor array of functionalized graphene quantum dots according to claim 1, characterized in that: In step S1, the method for synthesizing antibiotic-modified graphene quantum dots comprises: (1) Synthesis of spectinomycin-modified graphene quantum dots Anhydrous citric acid, thiourea and spectinomycin hydrochloride were added to water, stirred, and then the mixture was transferred to a Teflon-lined autoclave and heated at 140-170°C for 1-10 hours; (2) Synthesis of kanamycin-modified graphene quantum dots Anhydrous citric acid, L-cysteine ​​and kanamycin sulfate were added to water, stirred, and then heated in a Teflon-lined reactor at 150-200° C. for 5-15 hours; (3) Synthesis of polymyxin B-modified graphene quantum dots Anhydrous citric acid, sulfur powder, polymyxin B and ammonia water were added to water, stirred, and then heated at 150-200°C in a Teflon-lined autoclave for 10-18 hours.

4. The method for rapid pathogen detection and identification based on a machine learning enhanced fluorescence sensor array of functionalized graphene quantum dots according to claim 3, characterized in that: The fluorescence intensity of spectinomycin-modified graphene quantum dots, kanamycin-modified graphene quantum dots and polymyxin B-modified graphene quantum dots under the conditions of excitation wavelength of 300-440 nm and emission wavelength of 300-700 nm was measured.

5. The method for rapid pathogen detection and identification based on a machine learning enhanced fluorescence sensor array of functionalized graphene quantum dots according to claim 3, characterized in that: The spectinomycin-modified graphene quantum dots, kanamycin-modified graphene quantum dots and polymyxin B-modified graphene quantum dots were determined, and the presence of C, N, O, and S components in the spectinomycin-modified graphene quantum dots, kanamycin-modified graphene quantum dots and polymyxin B-modified graphene quantum dots as well as the Zeta potential of the three materials were determined by X-ray photoelectron spectroscopy.

6. The method for rapid pathogen detection and identification based on a machine learning enhanced fluorescence sensor array of functionalized graphene quantum dots according to claim 1, characterized in that: In step S3, the binding time is 3-8 min.

7. The method for rapid pathogen detection and identification based on a machine learning enhanced fluorescence sensor array of functionalized graphene quantum dots according to claim 1, characterized in that: In step S4, the fluorescence response data is the change in fluorescence intensity after the spectinomycin-modified graphene quantum dots, kanamycin-modified graphene quantum dots and polymyxin B-modified graphene quantum dots are mixed with pathogenic bacteria, wherein the formula for normalizing the obtained fluorescence response data is: (F0-F) / F0; Where F0 is the fluorescence intensity before adding pathogens, and F is the fluorescence intensity after adding pathogens.

8. The method for rapid pathogen detection and identification based on a machine learning enhanced fluorescence sensor array of functionalized graphene quantum dots according to claim 1, characterized in that: Pathogens include Salmonella typhimurium, Escherichia coli O157:H7 , Staphylococcus aureus, Pseudomonas aeruginosa, and Listeria monocytogenes.

9. The method for rapid pathogen detection and identification based on a machine learning enhanced fluorescence sensor array of functionalized graphene quantum dots according to claim 8, characterized in that: A total of thirteen repetitions were performed for each combination of graphene quantum dots and pathogens to construct a training data set, calculated as three graphene quantum dots times five pathogen species times thirteen repetitions.

Images