Method for photoelectric joint detection of nucleic acid protein based on machine learning assistance

Abstract

The invention relates to the technical field of biological detection, in particular to a method for photoelectric joint detection of nucleic acid protein based on machine learning assistance. The method comprises the following steps: firstly, connecting a photoelectric combined nucleic acid protein detection device with a test system; the method comprises the following steps: adding a nucleic acid protein sample to be detected into a sample pool in a photoelectric combined nucleic acid protein detection device to obtain D detection and G detection; performing data processing on the D measurement and the G measurement to obtain characteristic values of normalized electric sensing unit signal response and optical sensing unit signal response; training the machine learning model to obtain a classification surface I and a classification surface II; and finally, taking the normalized characteristic values of the signal response of the electric sensing unit and the signal response of the optical sensing unit as input and substituting into the classification surface I and the classification surface II to realize the interpretation result of the nucleic acid protein sample to be detected. The method provided by the invention can solve the problems of single acquired signal, low clinical accuracy and insufficient reliability of the existing molecular detection technology.

Description

Technical Field

[0001] The present invention relates to the field of biological detection technology, and in particular to a method for photoelectric combined detection of nucleic acid proteins based on machine learning assistance. Background Art

[0002] With the acceleration of the progress of digital medicine, it is necessary to develop fast and high-precision nucleic acid protein detection methods. In recent years, molecular detection methods have shown many advantages such as low cost, fast response, and simple operation. However, the existing molecular detection methods, including electrochemical sensors, mass spectrometers, surface Raman enhanced sensors, transistor sensors, and immunochromatographic sensors, generally monitor a single signal response and adopt a single judgment mode. This faces many challenges when dealing with unprocessed clinical samples, such as large loss of sensitivity in complex clinical sample detection, low detection accuracy, and complex signal judgment mode. Therefore, it is urgent to develop efficient and sensitive multi-signal sensors to achieve accurate and reliable nucleic acid protein detection. Summary of the invention

[0003] In order to solve the problems of existing molecular detection methods such as "single signal acquisition, low clinical accuracy and insufficient reliability", the purpose of the present invention is to provide a method for photoelectric combined detection of nucleic acid proteins based on machine learning.

[0004] Machine learning is a process of data mining through algorithms, and is now widely used in fields such as material design, artificial intelligence, and clinical analysis. In sensor detection applications, accurate decision classification can be made on input data by selecting appropriate machine learning parameters and algorithms. Furthermore, through mathematical dimension increase and visualization processing, machine learning can compare and analyze the acquired multi-signal responses, and ultimately achieve accurate detection of nucleic acid protein clinical samples by optoelectronic combined sensors.

[0005] The purpose of the present invention can be achieved by the following technical solutions:

[0006] The present invention provides a method for photoelectric combined detection of nucleic acid proteins based on machine learning, comprising the following steps:

[0007] (S1) connecting the photoelectric combined nucleic acid protein detection device to a test system;

[0008] (S2) adding the nucleic acid protein sample to be tested into the sample pool in the photoelectric combined detection nucleic acid protein device, and obtaining the signal response D of the electric sensing unit 测 and the signal response G of the light sensing unit 测 ;

[0009] (S3) Response D of the signal of the electric sensor unit obtained in step (S2) 测 and the signal response G of the light sensing unit 测Performing data processing to obtain characteristic values ​​of normalized electrical sensing unit signal response and optical sensing unit signal response;

[0010] (S4) training a machine learning model to obtain classification surface I and classification surface II;

[0011] (S5) Using the characteristic values ​​of the normalized electrical sensor unit signal response and the optical sensor unit signal response obtained in step (S3) as input and substituting them into the classification plane I and classification plane II obtained in step (S4) to realize the interpretation result of the nucleic acid protein sample to be tested.

[0012] In one embodiment of the present invention, in step (S1), the test system includes an electrical test system and an optical test system;

[0013] The optical testing system is a non-contact optical testing system, and the non-contact optical testing system is selected from one of an electrochemiluminescence instrument, a laser confocal microscope or a Raman spectrometer.

[0014] In one embodiment of the present invention, the electrical sensing unit of the optoelectronic combined detection device for nucleic acid and protein is connected to the electrical testing system, and the optical sensing unit is connected to the optical testing system.

[0015] In one embodiment of the present invention, the source electrode in the electrical sensing unit is connected to the positive electrode of the electrical testing system, and the drain electrode in the electrical sensing unit is connected to the negative electrode of the electrical testing system.

[0016] In one embodiment of the present invention, in step (S2), the nucleic acid protein is selected from one of the nucleic acid proteins of influenza virus, coronavirus, group B streptococcus, chlamydia, mycoplasma, bacteria or rickettsia.

[0017] In one embodiment of the present invention, in step (S2), the signal response D of the electric sensing unit 测 It is the steady-state current corresponding to the electrical test system after it stabilizes.

[0018] In one embodiment of the present invention, in step (S2), the signal response G of the light sensing unit 测 It is the corresponding photon count after the optical test system is stabilized.

[0019] In one embodiment of the present invention, in step (S3), a Z-score algorithm is used in the data processing process.

[0020] In one embodiment of the present invention, in step (S4), the process of training the machine learning model uses one of the partial least squares method, support vector machine algorithm, K nearest neighbor algorithm or logistic regression algorithm.

[0021] In one embodiment of the present invention, the process of training the machine learning model uses a logistic regression algorithm.

[0022] In one embodiment of the present invention, in step (S5), the interpretation result of the nucleic acid protein test sample is as follows:

[0023] When the characteristic values ​​of the normalized electrical sensing unit signal response and the optical sensing unit signal response belong to classification plane I, it is determined to be detected / positive, that is, the nucleic acid protein sample to be tested contains the pathogen to be tested;

[0024] When the characteristic values ​​of the normalized electrical sensor unit signal response and the optical sensor unit signal response do not belong to classification plane I, but belong to classification plane II, it is determined to be detected / positive, that is, the nucleic acid protein to be tested sample contains nucleic acid protein;

[0025] When the characteristic values ​​of the normalized electrical sensor unit signal response and the optical sensor unit signal response do not belong to classification plane I and do not belong to classification plane II, it is determined to be in the gray area and steps (S2) to (S5) are repeated.

[0026] Compared with the prior art, the present invention has the following beneficial effects:

[0027] The present invention combines the front-end photoelectric joint detection nucleic acid protein device with the back-end machine learning mathematical model. By optimizing the classification algorithm, it overcomes the problem of low clinical detection accuracy caused by the single recognition signal of the existing molecular detection method, and provides a new direction for the practical application of molecular detection in complex clinical samples. The auxiliary algorithm based on machine learning constructs classification surfaces for different nucleic acid proteins, providing a basis for accurate diagnosis of sensors, with a detection accuracy of more than 96.3%, which has potential social and economic value. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 This is a schematic diagram of the structure of the photoelectric combined detection device for nucleic acid and protein of Example 1;

[0029] Figure 2 This is a flow chart of the method for photoelectric combined detection of nucleic acid proteins based on machine learning assistance in Example 2;

[0030] Figure 3 The signal response D of the electric sensing unit is obtained in Example 2 测 Result graph;

[0031] Figure 4 The signal response G of the light sensing unit obtained in Example 2 测 Result graph;

[0032] Figure 5 This is a schematic diagram of the results of machine learning-assisted photoelectric combined detection of rhinovirus in Example 2;

[0033] Figure 6 This is a schematic diagram of the results of machine learning-assisted photoelectric combined detection of Mycobacterium tuberculosis in Example 3;

[0034] Figure 7 This is a schematic diagram of the results of machine learning-assisted photoelectric combined detection of group B streptococci in Example 4;

[0035] Figure 8 This is a schematic diagram of the detection results of the optoelectronic combined nucleic acid protein detection device in Example 5 using different dimensional classification methods;

[0036] Numbers in the figure: 1. insulating substrate; 2. source; 3. drain; 4. first connecting molecule; 5. second connecting molecule; 6. gate; 7. sample pool; 8. biological probe; 9. two-dimensional sensitive material. DETAILED DESCRIPTION

[0037] The present invention provides a method for photoelectric combined detection of nucleic acid proteins based on machine learning, comprising the following steps:

[0038] (S1) connecting the photoelectric combined nucleic acid protein detection device to a test system;

[0039] (S2) adding the nucleic acid protein sample to be tested into the sample pool in the photoelectric combined detection nucleic acid protein device, and obtaining the signal response D of the electric sensing unit 测 and the signal response G of the light sensing unit 测 ;

[0040] (S3) Response D of the signal of the electric sensor unit obtained in step (S2) 测 and the signal response G of the light sensing unit 测 Performing data processing to obtain characteristic values ​​of normalized electrical sensing unit signal response and optical sensing unit signal response;

[0041] (S4) training a machine learning model to obtain classification surface I and classification surface II;

[0042] (S5) Using the characteristic values ​​of the normalized electrical sensor unit signal response and the optical sensor unit signal response obtained in step (S3) as input and substituting them into the classification plane I and classification plane II obtained in step (S4) to realize the interpretation result of the nucleic acid protein sample to be tested.

[0043] In one embodiment of the present invention, in step (S1), the test system includes an electrical test system and an optical test system;

[0044] The optical testing system is a non-contact optical testing system, and the non-contact optical testing system is selected from one of an electrochemiluminescence instrument, a laser confocal microscope or a Raman spectrometer.

[0045] In one embodiment of the present invention, the electrical sensing unit of the optoelectronic combined detection device for nucleic acid and protein is connected to the electrical testing system, and the optical sensing unit is connected to the optical testing system.

[0046] In one embodiment of the present invention, the source electrode in the electrical sensing unit is connected to the positive electrode of the electrical testing system, and the drain electrode in the electrical sensing unit is connected to the negative electrode of the electrical testing system.

[0047] In one embodiment of the present invention, in step (S2), the nucleic acid protein is selected from one of the nucleic acid proteins of influenza virus, coronavirus, group B streptococcus, chlamydia, mycoplasma, bacteria or rickettsia.

[0048] In one embodiment of the present invention, in step (S2), the signal response D of the electric sensing unit 测 It is the steady-state current corresponding to the electrical test system after it stabilizes.

[0049] In one embodiment of the present invention, in step (S2), the signal response G of the light sensing unit 测 It is the corresponding photon count after the optical test system is stabilized.

[0050] In one embodiment of the present invention, in step (S3), a Z-score algorithm is used in the data processing process.

[0051] In one embodiment of the present invention, in step (S4), the process of training the machine learning model uses one of the partial least squares method, support vector machine algorithm, K nearest neighbor algorithm or logistic regression algorithm.

[0052] In one embodiment of the present invention, the process of training the machine learning model uses a logistic regression algorithm.

[0053] In one embodiment of the present invention, in step (S5), the interpretation result of the nucleic acid protein test sample is as follows:

[0054] When the characteristic values ​​of the normalized electrical sensing unit signal response and the optical sensing unit signal response belong to classification plane I, it is determined to be detected / positive, that is, the nucleic acid protein to be tested sample contains nucleic acid protein;

[0055] When the characteristic values ​​of the normalized electrical sensor unit signal response and the optical sensor unit signal response do not belong to classification plane I, but belong to classification plane II, it is determined to be undetected / negative, that is, the nucleic acid protein to be tested sample does not contain nucleic acid protein;

[0056] When the characteristic values ​​of the normalized electrical sensor unit signal response and the optical sensor unit signal response do not belong to classification plane I and do not belong to classification plane II, it is determined to be in the gray area and steps (S2) to (S5) are repeated.

[0057] The present invention is described in detail below with reference to the accompanying drawings and specific embodiments.

[0058] The following detection methods are not intended for disease diagnosis, but simply for the presence of rhinovirus (HRV), tuberculosis (TB), group B streptococcus (GBS), new coronavirus, and influenza A virus (H1N1) in the sample to be tested.

[0059] The rhinovirus (HRV) test samples used in the detection method of the present invention include nasal swab and throat swab samples, the TB test samples used are lower respiratory tract samples (deep cough sputum, bronchial lavage fluid, alveolar lavage fluid) and serum samples, and the group B streptococcus (GBS) test samples used are lower vaginal and anal swab samples. The specific acquisition method is as follows:

[0060] (a) Nasal swab and throat swab: The specific steps for collecting and processing samples to be tested should refer to the "National Influenza Surveillance Technical Guidelines (2017 Edition)" of the China Center for Disease Control and Prevention, the "Technical Specifications for 10-in-1 Mixed Sampling of Novel Coronavirus Nucleic Acid Testing" of the National Health Commission of the People's Republic of China, and the "Technical Guidelines for Dilution and Mixed Sampling Testing of Novel Coronavirus Nucleic Acid Screening" of the National Health Commission of the People's Republic of China;

[0061] (b) Deep cough sputum: ask the patient to cough deeply and collect the sputum into a 50-mL screw-capped plastic tube containing 3 mL of sampling solution;

[0062] (c) Bronchial lavage fluid: insert the collector head into the trachea (about 30 cm deep) from the nostril or tracheal intubation, inject 5 ml of normal saline, turn on negative pressure, rotate the collector head and slowly withdraw it. Collect the extracted mucus and rinse the collector once with the sampling fluid (you can also use a pediatric catheter connected to a 50 ml syringe to replace the collection);

[0063] (d) Bronchoalveolar lavage fluid: After local anesthesia, insert a fiberoptic bronchoscope through the mouth or nose through the pharynx into the branch of the right middle lobe or the left lingular segment of the lung, insert its tip into the opening of the bronchial branch, and slowly add sterile saline through the tracheal biopsy hole, 30-50 ml each time, with a total volume of 100-250 ml, which should not exceed 300 ml;

[0064] (e) Lower vaginal and anal swabs: Without using a vaginal speculum, use a swab to collect samples in the lower 1 / 3 of the vagina, and then use the same swab to collect samples in the rectum through the rectal sphincter. For specific methods, please refer to the Chinese Medical Association Perinatal Medicine Branch, Chinese Medical Association Obstetrics and Gynecology Branch Obstetrics Group. Prevention of Perinatal Group B Streptococcal Disease (China) Expert Consensus [J] (Chinese Journal of Perinatal Medicine, 2021, 24, 561).

[0065] Example 1

[0066] This embodiment provides a photoelectric combined detection device for nucleic acid protein, such as Figure 1 As shown, the photoelectric combined detection nucleic acid protein device includes an insulating substrate 1, a source 2, a drain 3, a connecting molecule (a first connecting molecule 4 and a second connecting molecule 5), a gate 6, a sample pool 7, a biological probe 8 and a two-dimensional sensitive material 9, wherein the source 2, the drain 3 and the gate 6 are located above the insulating substrate 1, wherein the source 2 and the drain 3 are arranged in parallel, and the gate 6 is located opposite to the source 2 and the drain 3; a sample pool 7 is arranged above the source 2, the drain 3 and the gate 6; the two-dimensional sensitive material 9 is located at the bottom of the source 2 and the drain 3 close to the end surface, and is located on the upper surface of the insulating substrate 1; a first connecting molecule 4 with a luminescent group is arranged above the two-dimensional sensitive material 9; the second connecting molecule 5 is connected to the end of the gate 6; the upper surface of the connecting molecule (the first connecting molecule 4 and the second connecting molecule 5) is connected to the biological probe 8; on the one hand, the insulating substrate 1, the source 2, the drain 3, the gate 6, the first connecting molecule 4, the biological probe 8 and the two-dimensional sensitive material 9 constitute an electrical sensing unit; on the other hand, the insulating substrate 1, the second connecting molecule 5, the biological probe 8 and the two-dimensional sensitive material 9 constitute a light sensing unit.

[0067] Among them, the insulating substrate 1 is a plastic film with a transmittance greater than 90%, and the plastic film is selected from one or more of polyester resin, polyimide or polymethyl methacrylate; the source electrode 2 is a gold source electrode, the drain electrode 3 is a gold drain electrode, and the gate electrode 6 is a gold gate; the first connecting molecule 4 is selected from one of gold nanoparticles or carbon quantum dots; the second connecting molecule 5 is selected from one of coupling activators or gold nanoparticles; the biological probe 8 is selected from one of antibody molecules, nucleic acid probes or antigen molecules; the two-dimensional sensitive material 9 is a two-dimensional sensitive material film obtained by chemical deposition, mechanical stripping or liquid phase stripping.

[0068] Example 2

[0069] This embodiment provides a method for photoelectric combined detection of nucleic acid proteins based on machine learning.

[0070] In this embodiment, the nucleic acid protein is HRV, and the specific steps are as follows (eg Figure 2 shown):

[0071] (S10) Transistor combined with photoelectric sensing system construction:

[0072] (S101) Metal electrodes (50 nanometers of gold) are processed on a PI film with a thickness of 2 millimeters by means of printed electronics technology. The electrodes include a gold gate, a gold source, and a gold drain. The gold source is a current input terminal, and the gold drain is a current output terminal.

[0073] (S102) spin coating 8 wt.% of polymethyl methacrylate (PMMA) on the two-dimensional sensitive material grown on the metal substrate, and transferring the two-dimensional sensitive material / PMMA film to a quartz substrate with a metal electrode by an electrochemical method and connecting it between a gold source electrode and a gold drain electrode; soaking it in acetone for 2 hours, then washing it with isopropanol and deionized water respectively, and etching the two-dimensional sensitive material / PMMA film into a 150×20 micron rectangular shape by photolithography technology to obtain a device to be modified;

[0074] (S103) Modifying the second linker molecule in the gold gate region of the device to be modified: Soaking the gold gate in a 50 μg / ml Staphylococcus aureus solution for 1 hour to modify the gold nanoparticles.

[0075] Modify the first connecting molecule with a luminescent group in the channel region of the two-dimensional sensitive material: soak the two-dimensional sensitive material / PMMA film in a 5 mmol / L PASE solution for 2 hours; then soak it in a 25 mmol / L anthocyanin fluorescent dye (Cy3 dye, CAS#146368-13-0) solution in the dark for 2 hours.

[0076] (S104) The gold nanoparticles and the two-dimensional sensitive material / PMMA film on the device to be modified are immersed in a 5 μg / ml MPT64 antibody solution for 6 hours, and then washed with a 1×PBS buffer solution to obtain a pre-treated device.

[0077] (S105) 194 type polydimethylsiloxane (PDMS) is mixed at 194A:194B=1:10 (wt.%), and after removing air by vacuum, the mixture is baked at 70°C for 30 minutes to obtain PDMS. A hole is punched in the PDMS to obtain a sample pool, which is placed above a pretreatment device to obtain a photoelectric combined nucleic acid protein detection device.

[0078] (S20) Connecting the gold source electrode, gold gate electrode and gold drain electrode in the photoelectric combined detection nucleic acid protein device obtained in step (S10) to the electrical test system via crocodile clips, wherein the gold source electrode is connected to the positive electrode of the electrical test system, and the gold drain electrode is connected to the negative electrode of the electrical test system; connecting the optical sensing unit to the 20× optical path of the laser confocal microscope (laser beam 532 nanometers); adding the HRV sample to be tested into the sample pool, and obtaining the signal response D of the electrical sensing unit 测 ( Figure 3 ) and the signal response G of the light sensing unit 测 ( Figure 4 ).

[0079] (S30) Using the Z-Score algorithm of SPSS27 to obtain the signal response D of the electric sensor unit 测 and the signal response G of the light sensing unit 测 Carry out data processing;

[0080] (S40) After step (S30) is completed, feature extraction is performed on the signal response of the electrical sensor unit and the signal response of the optical sensor unit to obtain feature values ​​of the normalized signal response of the electrical sensor unit and the signal response of the optical sensor unit;

[0081] (S50) Using Python 3.1 to train a logistic regression algorithm to model the data of the eigenvalues ​​of the normalized electrical sensor unit signal response and the optical sensor unit signal response obtained in step (S40), and obtain classification plane I and classification plane II ( Figure 5 );

[0082] (S60) Substituting the characteristic values ​​of the normalized electrical sensor unit signal response and the optical sensor unit signal response into the classification surface I as input;

[0083] Interpretation results:

[0084] When the characteristic values ​​of the normalized electrical sensor unit signal response and the optical sensor unit signal response belong to classification plane I, it is determined to be detected / positive, that is, the sample to be tested contains HRV;

[0085] When the characteristic values ​​of the normalized electrical sensor unit signal response and the optical sensor unit signal response do not belong to classification plane I, but belong to classification plane II, it is determined to be undetected / negative, that is, the sample to be tested does not contain HRV;

[0086] When the characteristic values ​​of the normalized electrical sensor unit signal response and the optical sensor unit signal response do not belong to classification plane I and do not belong to classification plane II, it is determined to be in the gray area and steps (S20) to (S50) are repeated.

[0087] The accuracy of HRV detection using the method of this embodiment is 99.8%.

[0088] The HRV sample to be tested in this embodiment is provided by the Public Health Clinical Center affiliated to Fudan University, and the HRV concentration contained therein is calibrated to 11 copies per milliliter. The biological probes used in this embodiment include: HRV monoclonal antibody (60299-1-1g, Proteintech), HRV polyclonal antibody (10831-1-AP, Proteintech) and HRV nucleic acid probe (SEQ ID NO.1: 5'-TCCTCCGGCCCCTGAATGYGGCTA-3').

[0089] Example 3

[0090] This embodiment provides a method for photoelectric combined detection of nucleic acid proteins based on machine learning.

[0091] In this embodiment, the nucleic acid protein is TB; the difference between this embodiment and embodiment 2 is that:

[0092] In step (S101), the insulating substrate used in this embodiment is a polyester resin film (thickness 1000 microns);

[0093] In step (S103), the TB antibodies used in this example are anti-MPT64 (Abcam, ab193435) and anti-CFP10 (Invitrogen, MA5-18220);

[0094] In step (S20), the samples to be tested in this embodiment are 173 TB clinical samples. The specific test results are as follows: Figure 6 As shown:

[0095] The accuracy of machine learning-assisted photoelectric combined detection of TB reached 98.9%, showing excellent detection performance, which can reflect the value of the method described in the present invention.

[0096] Example 4

[0097] This embodiment provides a method for photoelectric combined detection of nucleic acid proteins based on machine learning.

[0098] The nucleic acid protein in this embodiment is GBS; the difference between this embodiment and embodiment 2 is that:

[0099] In step (S101), the insulating substrate used in this embodiment is a polymethyl methacrylate film (thickness 1000 microns);

[0100] In step (S103), the GBS nucleic acid probe sequence used in this embodiment is: 5'-GGTGCATTGTTATTTTCACCA-3' (SEQ ID NO. 2);

[0101] In step (S20), the specimens to be tested in this embodiment are 169 GBS clinical samples, and the specific test results are as follows: Figure 7 As shown:

[0102] The accuracy of machine learning-assisted optoelectronic combined detection of GBS reached 96.3%, showing excellent detection performance, which can reflect the value of the method described in the present invention.

[0103] Example 5

[0104] This embodiment provides a method for detecting nucleic acid proteins using a photoelectric combined nucleic acid protein detection device using different dimensional classification methods.

[0105] In this embodiment, the nucleic acid proteins are HRV, TB and GBS respectively; the difference between this embodiment and embodiment 2 is that:

[0106] In step (S103), the TB antibodies used in this example are anti-MPT64 (Abcam, ab193435) and anti-CFP10 (Invitrogen, MA5-18220), and the GBS nucleic acid probe sequence used is: 5'-GGTGCATTGTTATTTTCACCA-3' (SEQ ID NO. 2);

[0107] In step (S20), the samples to be tested in this embodiment are 159 HRV clinical samples, 173 TB clinical samples and 169 GBS clinical samples;

[0108] In step (S50), in this embodiment, different dimensional (1D, 2D and 3D) algorithms are used to classify the detection results, so as to evaluate the detection accuracy. The specific test results are as follows: Figure 8 As shown:

[0109] The accuracy of machine learning-assisted optoelectronic combined detection of nucleic acid proteins based on a 3D algorithm is greater than 96.3%, while the detection accuracy of the low-dimensional algorithm is lower (1D algorithm is less than 82%, 2D algorithm is less than 95%), demonstrating the excellent detection performance of the detection method of the present invention and reflecting the value of the method described in the present invention.

[0110] Example 6

[0111] This embodiment provides a method for photoelectric combined detection of nucleic acid proteins based on machine learning.

[0112] In this embodiment, the nucleic acid protein is a novel coronavirus; the difference between this embodiment and embodiment 2 is that:

[0113] In step (S101), the insulating substrate used in this embodiment is a polymethyl methacrylate film (thickness 1000 microns);

[0114] In step (S103), the new coronavirus antibodies used in this embodiment include SARS-CoV-2 Spike S1 antibody (40150-R007, SinoBiological) and CR3022 antibody (ab273073, Abcam), and the new coronavirus nucleic acid probe sequence used is 5'-CCATAACCTTTCCACATACCGCAGACGG-3' (SEQ ID NO.3).

[0115] In step (S20), the specimens to be tested in this embodiment are 17 clinical samples of the new coronavirus. The specific test results are shown in Table 1:

[0116] Table 1

[0117]

[0118]

[0119] In the clinical testing of 17 new coronavirus samples in this example, the comprehensive accuracy of machine learning-assisted optoelectronic combined detection of the new coronavirus was 100%.

[0120] Example 7

[0121] The method described in the specific implementation manner is used to perform machine learning-assisted photoelectric combined detection of influenza A virus (H1N1). The difference between this embodiment and embodiment 2 is that:

[0122] In step (S103), the H1N1 nucleic acid probe sequences used in this embodiment are 5'-TTTTTGTCCTCGCTC-3' (H1N1-1: SEQ ID NO. 4) and 5'-TTTTTACACAAATCCTAAAATTCCC-3' (H1N1-2: SEQ ID NO. 5).

[0123] In step (S20), the samples to be tested in this embodiment are 21 H1N1 clinical samples. The specific test results are shown in Table 1:

[0124] Table 2

[0125]

[0126]

[0127] In the clinical testing of 21 novel coronavirus samples in this example, the comprehensive accuracy of machine learning-assisted photoelectric combined detection of H1N1 was 100%.

[0128] The above description of the embodiments is to facilitate the understanding and use of the invention by those skilled in the art. It is obvious that those skilled in the art can easily make various modifications to these embodiments and apply the general principles described herein to other embodiments without creative work. Therefore, the present invention is not limited to the above embodiments, and improvements and modifications made by those skilled in the art based on the explanation of the present invention without departing from the scope of the present invention should be within the scope of protection of the present invention.

Claims

1. A method for photoelectric combined detection of nucleic acid proteins based on machine learning, characterized in that: The following steps are involved: (S1) connecting the photoelectric combined nucleic acid protein detection device to a test system; (S2) adding the nucleic acid protein sample to be tested into the sample pool in the photoelectric combined detection nucleic acid protein device, and obtaining the signal response D of the electric sensing unit 测 and the signal response G of the light sensing unit 测 ; (S3) Response D of the signal of the electric sensing unit obtained in step (S2) 测 and the signal response G of the light sensing unit 测 Performing data processing to obtain characteristic values ​​of normalized electrical sensing unit signal response and optical sensing unit signal response; (S4) training a machine learning model to obtain classification surface I and classification surface II; (S5) Using the characteristic values ​​of the normalized electrical sensor unit signal response and the optical sensor unit signal response obtained in step (S3) as input and substituting them into the classification plane I and classification plane II obtained in step (S4) to realize the interpretation result of the nucleic acid protein sample to be tested.

2. The method of claim 1, wherein the method comprises: In step (S1), the test system includes an electrical test system and an optical test system.

3. The method of claim 2, wherein the method comprises: The electrical sensing unit of the optoelectronic combined nucleic acid protein detection device is connected to the electrical testing system, and the optical sensing unit is connected to the optical testing system.

4. The method of claim 1, wherein: In step (S2), the nucleic acid protein is selected from one of the nucleic acid proteins of influenza virus, coronavirus, group B streptococcus, chlamydia, mycoplasma, bacteria or rickettsia.

5. The method of claim 1, wherein: In step (S2), the signal response D of the electric sensing unit 测 It is the steady-state current corresponding to the electrical test system after it stabilizes.

6. The method of claim 1, wherein: In step (S2), the signal response G of the light sensing unit 测 It is the corresponding photon count after the optical test system is stabilized.

7. The method of claim 1, wherein: In step (S3), the Z-score algorithm is used in the data processing process.

8. The method of claim 1, wherein: In step (S4), the process of training the machine learning model uses one of the partial least squares method, support vector machine algorithm, K nearest neighbor algorithm or logistic regression algorithm.

9. The method of claim 8, wherein: The process of training the machine learning model uses the logistic regression algorithm.

10. The method of claim 1, wherein: In step (S5), the results of the nucleic acid protein test sample are as follows: When the characteristic values ​​of the normalized electrical sensing unit signal response and the optical sensing unit signal response belong to classification plane I, it is determined to be detected / positive, that is, the nucleic acid protein to be tested sample contains nucleic acid protein; When the characteristic values ​​of the normalized electrical sensor unit signal response and the optical sensor unit signal response do not belong to classification plane I, but belong to classification plane II, it is determined to be undetected / negative, that is, the nucleic acid protein to be tested sample does not contain nucleic acid protein; When the characteristic values ​​of the normalized electrical sensor unit signal response and the optical sensor unit signal response do not belong to classification plane I and do not belong to classification plane II, it is determined to be in the gray area and steps (S2) to (S5) are repeated.

Images