New crown growing typing biomarker group and application thereof, typing product and typing method
A biomarker panel with specific biomarkers and a neural network model accurately type long-COVID, addressing diagnostic challenges by identifying distinct subtypes for targeted interventions and treatment.
Patent Information
- Application Number
- CN202411386468.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Priority Date
- 2024-06-20
- Filing Date
- 2024-09-29
- Publication Date
- 2024-12-06
AI Technical Summary
Current diagnostic methods for long-COVID are inadequate, failing to accurately distinguish and diagnose different subtypes of the disease, which complicates targeted intervention and treatment.
A biomarker panel comprising specific biomarkers (FPGT, AMDHD2, PSME4, TSC2, CBX6, ACACB, HNRNPUL1, DGKH, DDX20, RBBP4, PCNA, CDK11B, CRH, LPCAT2) is used to detect protein or mRNA expression levels, combined with a multivariate neural network model for accurate long-COVID typing, identifying types such as nervous, cardiovascular, skeletal muscle, and multi-system composite.
The biomarker panel and neural network model provide high sensitivity and specificity for long-COVID typing, enabling targeted interventions and early pathological mechanism studies, improving diagnostic accuracy and treatment efficacy.
Smart Images

Figure CN119082294A_ABST
Abstract
Description
Technical Field
[0001] The present disclosure relates to the technical field of long-COVID-19 typing, and specifically, to a biomarker group for long-COVID-19 typing and its application, a long-COVID-19 typing product and a typing method. Background Art
[0002] According to the definition of the World Health Organization, long COVID refers to the continuation or emergence of new symptoms 3 months after COVID-19 infection, including fatigue, shortness of breath and cognitive dysfunction, which lasts for at least 2 months. According to surveys, the incidence of long COVID ranges from 1.9% to 20%. As of April 2024, nearly 800 million cases of COVID-19 infection have been reported to the World Health Organization. It is conservatively estimated that at least tens of millions of people in the world have experienced long COVID. Long COVID is a multi-manifestation disease that can affect multiple systems of the body, including respiratory, nervous, and circulatory systems. According to statistics, there are more than 200 symptoms of long COVID.
[0003] The huge number of long-COVID patients has placed a huge burden on society and the medical and health care system. The first step to alleviate and treat long-COVID patients is to correctly diagnose long-COVID. However, there is currently no specific diagnostic method for long-COVID. Clinicians may use blood tests, electrocardiograms, imaging examinations, psychological assessments and other means to comprehensively evaluate the body state of long-COVID patients to find possible causes. However, routine test results of long-COVID patients are usually within the normal range, and abnormal test results may not be related to long-COVID symptoms. At the same time, current studies have found that the mechanisms of long-COVID may be different between different systems, and conventional auxiliary examinations cannot effectively distinguish and diagnose different subtypes of long-COVID.
[0004] It should be noted that the information disclosed in the above background technology section is only used to enhance the understanding of the background of the present disclosure, and therefore may include information that does not constitute the prior art known to ordinary technicians in the field. Summary of the invention
[0005] The purpose of the present disclosure is to overcome the deficiencies of the above-mentioned prior art, and to provide a biomarker group for long-COVID typing and its application, typing products and typing methods, so as to achieve long-COVID typing of the samples to be tested through the corresponding biomarker detection results of the samples to be tested.
[0006] According to one aspect of the present disclosure, a biomarker panel for typing long-term novel coronavirus is provided, the biomarker panel comprising:
[0007] at least one of the biomarkers FPGT, AMDHD2 of the first biomarker subgroup;
[0008] at least one of the biomarkers PSME4, TSC2, CBX6, ACACB of the second biomarker subgroup;
[0009] at least one of the biomarkers HNRNPUL1 and DGKH of the third biomarker subgroup;
[0010] at least one of the biomarkers DDX20, RBBP4, and PCNA of the fourth biomarker subgroup;
[0011] and, at least one of the biomarkers CDK11B, CRH, LPCAT2 of the fifth biomarker subgroup.
[0012] In an exemplary embodiment of the present disclosure, the biomarker panel includes AMDHD2 protein, ACACB protein, DGKH protein, DDX20 protein and CDK11B protein.
[0013] In an exemplary embodiment of the present disclosure, the biomarker panel includes mRNA of AMDHD2 protein, mRNA of ACACB protein, mRNA of DGKH protein, mRNA of DDX20 protein, and mRNA of CDK11B protein.
[0014] In an exemplary embodiment of the present disclosure, the biomarker panel includes FPGT protein, CBX6 protein, DGKH protein, RBBP4 protein and CRH protein.
[0015] In an exemplary embodiment of the present disclosure, the biomarker panel includes mRNA of FPGT protein, mRNA of CBX6 protein, mRNA of DGKH protein, mRNA of RBBP4 protein and mRNA of CRH protein.
[0016] According to another aspect of the present disclosure, there is provided an application of the above-mentioned biomarker group in the preparation of long-term new coronary typing products.
[0017] According to another aspect of the present disclosure, a long-COVID typing product is provided, which includes reagents for detecting each biomarker in the above-mentioned biomarker group.
[0018] In an exemplary embodiment of the present disclosure, the product is a kit, a detection reagent or a biochip.
[0019] According to another aspect of the present disclosure, a long-term new crown typing method is provided, the method comprising the following steps:
[0020] S1. Obtain a sample to be tested, wherein the sample to be tested is taken from a patient with COVID-19;
[0021] S2. Use the above-mentioned long new crown typing product to detect the protein expression levels of the first biomarker, the second biomarker, the third biomarker, the fourth biomarker, and the fifth biomarker in the sample to be tested to obtain the protein expression results of the sample to be tested;
[0022] wherein the first biomarker is selected from at least one biomarker in the first biomarker subgroup; the second biomarker is selected from at least one biomarker in the second biomarker subgroup; the third biomarker is selected from at least one biomarker in the third biomarker subgroup; the fourth biomarker is selected from at least one biomarker in the fourth biomarker subgroup; and the fifth biomarker is selected from at least one biomarker in the fifth biomarker subgroup;
[0023] S3, inputting the protein expression results of the sample to be tested obtained in step S2 into the long-term new crown multi-neural network language diagnosis model to obtain the output results of the long-term new crown multi-neural network language diagnosis model;
[0024] If the output result is the first biomarker, the sample to be tested is of the nervous system type; if the output result is the second biomarker, the sample to be tested is of the composite type of the cardiovascular system and the nervous system; if the output result is the third biomarker, the sample to be tested is of the composite type of the skeletal muscle system and the whole body nonspecific system; if the output result is the fourth biomarker, the sample to be tested is of the composite type of the cardiovascular system and the respiratory system; if the output result is the fifth biomarker, the sample to be tested is of the multi-system composite type, and the multi-system composite type includes the composite type of at least four systems including the nervous system, the circulatory system, the respiratory system, the skeletal muscle system, and the whole body nonspecific system.
[0025] In an exemplary embodiment of the present disclosure, the long new crown typing method comprises the following steps:
[0026] S1. Obtain a sample to be tested, wherein the sample to be tested is taken from a patient with COVID-19;
[0027] S2. Use the above-mentioned long new crown typing product to detect the expression levels of AMDHD2 protein, ACACB protein, DGKH protein, DDX20 protein and CDK11B protein in the sample to be tested to obtain the protein expression results of the sample to be tested;
[0028] S3. Input the protein expression results of the sample to be tested obtained in step S2 into the long-term new crown multivariate neural network language diagnosis model to obtain the output result of the long-term new crown multivariate neural network language diagnosis model; if the output result is AMDHD2 protein, the sample to be tested is a nervous system type; if the output result is ACACB protein, the sample to be tested is a composite type of cardiovascular system and nervous system; if the output result is DDX20 protein, the sample to be tested is a composite type of cardiovascular system and respiratory system; if the output result is DGKH protein, the sample to be tested is a composite type of skeletal muscle system and systemic nonspecific system; if the output result is CDK11B protein, the sample to be tested is a multi-system composite type.
[0029] In an exemplary embodiment of the present disclosure, the long new crown typing method comprises the following steps:
[0030] S1. Obtain a sample to be tested, wherein the sample to be tested is taken from a patient with COVID-19;
[0031] S2. Use the above-mentioned long new crown typing product to detect the expression levels of FPGT protein, CBX6 protein, DGKH protein, RBBP4 protein and CRH protein in the sample to be tested, and obtain the protein expression results of the sample to be tested;
[0032] S3. Input the protein expression results of the sample to be tested obtained in step S2 into the long-term new crown multi-neural network language diagnosis model to obtain the output result of the long-term new crown multi-neural network language diagnosis model; if the output result is FPGT protein, the sample to be tested is a nervous system type; if the output result is CBX6 protein, the sample to be tested is a composite type of cardiovascular system and nervous system; if the output result is RBBP4 protein, the sample to be tested is a composite type of cardiovascular system and respiratory system; if the output result is DGKH protein, the sample to be tested is a composite type of skeletal muscle system and systemic nonspecific system; if the output result is CRH protein, the sample to be tested is a multi-system composite type.
[0033] In an exemplary embodiment of the present disclosure, the construction process of the long-term new crown multi-neural network language diagnosis model in step S3 is:
[0034] Samples of long-COVID subjects were obtained, and the long-COVID subjects were divided into five experimental groups, including: a cardiovascular system and respiratory system composite long-COVID group, a cardiovascular system and nervous system composite long-COVID group, a nervous system type long-COVID group, a skeletal muscle system and systemic nonspecific system composite long-COVID group, and a multi-system composite long-COVID group;
[0035] Obtaining the protein expression levels of the first biomarker, the second biomarker, the third biomarker, the fourth biomarker and the fifth biomarker of each sample in each experimental group; the protein expression levels of the first biomarker, the second biomarker, the third biomarker, the fourth biomarker and the fifth biomarker of each sample in each experimental group constitute a data set;
[0036] 70% to 80% of the data in the data set were randomly selected as the training set, and the training set was input into the multivariate neural network language diagnosis model for iterative training to obtain the long-term new crown multivariate neural network language diagnosis model.
[0037] The biomarker group provided by the present invention has high sensitivity and specificity, and can accurately detect the typing of long COVID-19 to improve the accuracy of long COVID-19 typing screening; at the same time, it is convenient for targeted intervention of different types of long COVID-19.
[0038] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS
[0039] The accompanying drawings herein are incorporated into the specification and constitute a part of the specification, illustrate embodiments consistent with the present disclosure, and together with the specification are used to explain the principles of the present disclosure. Obviously, the accompanying drawings described below are only some embodiments of the present disclosure, and for ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without creative work.
[0040] Figure 1 This is a schematic diagram of sample-by-sample similarity matrix results of similarity network fusion in one embodiment of the present disclosure.
[0041] Figure 2 In one embodiment of the present disclosure, Figure 1 A magnified schematic diagram of local A.
[0042] Figure 3 This is a schematic diagram of the enriched signal pathway results of proteome-transcriptome joint analysis in one embodiment of the present disclosure.
[0043] Figure 4 This is a schematic diagram of significant marker proteins screened by the MULTI group in one embodiment of the present disclosure.
[0044] Figure 5 This is a schematic diagram of significant marker proteins screened by the NEU group in one embodiment of the present disclosure.
[0045] Figure 6 This is a schematic diagram of significant marker proteins screened by the MSK+SYST group in one embodiment of the present disclosure.
[0046] Figure 7 This is a schematic diagram of significant marker proteins screened by the CACRB group in one embodiment of the present disclosure.
[0047] Figure 8 This is a schematic diagram of significant marker proteins screened by the CAPM group in one embodiment of the present disclosure.
[0048] Fig. 9 Schematic diagram of ROC curve results of CDK11B protein, AMDHD2 protein, ACACB protein, DGKH protein and DDX20 protein as biomarker group in one embodiment of the present disclosure; wherein, (Ⅰ) is the ROC curve diagram of CDK11B protein selected in MULTI group, (Ⅱ) is the ROC curve diagram of AMDHD2 protein selected in NEU group, (Ⅲ) is the ROC curve diagram of ACACB protein selected in CACRB group, (Ⅳ) is the ROC curve diagram of DGKH protein selected in MSK+SYST group, and (Ⅴ) is the ROC curve diagram of DDX20 protein selected in CAPM group.
[0049] Fig.10 The present invention discloses an embodiment of a schematic diagram of the expression level comparison of CDK11B protein, AMDHD2 protein, ACACB protein, DGKH protein and DDX20 protein in each group; wherein, (Ⅰ) is a schematic diagram of the expression level of CDK11B protein in each group, (Ⅱ) is a schematic diagram of the expression level of AMDHD2 protein in each group, (Ⅲ) is a schematic diagram of the expression level of ACACB protein in each group, (Ⅳ) is a schematic diagram of the expression level of DGKH protein in each group, and (Ⅴ) is a schematic diagram of the expression level of DDX20 protein in each group.
[0050] Fig.11 In one embodiment of the present invention, CRH protein, FPGT protein, CBX6 protein, DGKH protein and RBBP4 protein are used as ROC curve results of the biomarker group; wherein, (Ⅰ) is the ROC curve diagram of CRH protein selected in the MULTI group, (Ⅱ) is the ROC curve diagram of FPGT protein selected in the NEU group, (Ⅲ) is the ROC curve diagram of CBX6 protein selected in the CACRB group, (Ⅳ) is the ROC curve diagram of DGKH protein selected in the MSK+SYST group, and (Ⅴ) is the ROC curve diagram of RBBP4 protein selected in the CAPM group. DETAILED DESCRIPTION
[0051] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in many forms and should not be construed as limited to the embodiments set forth herein; rather, these embodiments are provided so that this disclosure will be comprehensive and complete and will fully convey the concept of the example embodiments to those skilled in the art.
[0052] Unless otherwise defined, all terms and phrases used herein include the meanings that the terms and phrases have in the art, unless explicitly stated to the contrary or clearly apparent from the context in which the terms and phrases are used. Although any methods and materials similar or equivalent to those described herein can be used in the practice or testing of the present invention, specific methods and materials are now described.
[0053] The long-term COVID-19 symptoms that last for several months after COVID-19 infection have a significant impact on the patient's physical health and quality of life, while also increasing the burden on society and the medical and health systems. The different symptoms of long COVID-19 also increase the difficulty of intervention and treatment of long COVID-19.
[0054] To solve the above problems, the disclosed embodiment provides a biomarker group for long new crown typing, the biomarker group including: at least one of the biomarkers FPGT and AMDHD2 in the first biomarker subgroup; and at least one of the biomarkers PSME4, TSC2, CBX6, and ACACB in the second biomarker subgroup; and at least one of the biomarkers HNRNPUL1 and DGKH in the third biomarker subgroup; and at least one of the biomarkers DDX20, RBBP4, and PCNA in the fourth biomarker subgroup; and at least one of the biomarkers CDK11B, CRH, and LPCAT2 in the fifth biomarker subgroup. In this way, targeted intervention can be carried out on long new crown patients according to different long new crown typing. At the same time, a biomarker group is provided for the study of the pathological mechanism of each typing of long new crown, in order to provide research ideas for early intervention of long new crown through pathological mechanism.
[0055] In one embodiment of the present disclosure, the biomarker group includes AMDHD2 protein, ACACB protein, DGKH protein, DDX20 protein and CDK11B protein. Based on the above five protein markers, patients with long new crown can be typed. For example, the above five protein markers in the sample to be tested are detected. If the test result is AMDHD2 protein, the sample to be tested is a nervous system type long new crown. If the test result is ACACB protein, the sample to be tested is a cardiovascular system and nervous system composite long new crown. If the test result is DGKH protein, the sample to be tested is a skeletal muscle system and systemic nonspecific system composite long new crown. If the test result is DDX20 protein, the sample to be tested is a cardiovascular system and respiratory system composite long new crown. If the test result is CDK11B protein, the sample to be tested is a multi-system composite long new crown, wherein the multi-system composite is a composite of at least four systems including the nervous system, circulatory system, respiratory system, skeletal muscle system, and systemic nonspecific system. In this way, targeted intervention can be carried out on patients with long-term new coronary pneumonia according to different long-term new coronary pneumonia classifications. At the same time, five protein markers are provided for the study of the pathological mechanisms of various types of long-term new coronary pneumonia, in order to provide research ideas for the early intervention of long-term new coronary pneumonia through pathological mechanisms.
[0056] In one embodiment of the present disclosure, the biomarker group includes mRNA of AMDHD2 protein, mRNA of ACACB protein, mRNA of DGKH protein, mRNA of DDX20 protein and mRNA of CDK11B protein. In this way, the level of messenger RNA in the gene transcription process can be detected to type the long-term new crown patients.
[0057] That is to say, when typing the long new crown, the five protein markers in the sample to be tested can be detected to confirm the long new crown typing of the sample to be tested; the five mRNA markers in the sample to be tested can also be detected to confirm the long new crown typing of the sample to be tested.
[0058] In one embodiment of the present disclosure, the biomarker panel includes FPGT protein, CBX6 protein, DGKH protein, RBBP4 protein and CRH protein.
[0059] In one embodiment of the present disclosure, the biomarker panel includes mRNA of FPGT protein, mRNA of CBX6 protein, mRNA of DGKH protein, mRNA of RBBP4 protein and mRNA of CRH protein.
[0060] The disclosed embodiments provide an application of a biomarker group in the preparation of a long-term new crown typing product, including but not limited to, using the biomarker group as a target to prepare a long-term new crown typing product.
[0061] An embodiment of the present disclosure provides a long new crown typing product, which includes a reagent for detecting the expression level of each protein in the biomarker group.
[0062] In one embodiment of the present disclosure, the biomarker group may be AMDHD2 protein, ACACB protein, DGKH protein, DDX20 protein and CDK11B protein. In another example, the biomarker may also be FPGT protein, CBX6 protein, DGKH protein, RBBP4 protein and CRH protein.
[0063] Furthermore, the long-term new crown typing product includes but is not limited to a kit, a detection reagent or a biochip. Exemplarily, the kit can be an enzyme-linked immunosorbent assay kit (ELISA kit), which contains capture antibodies and detection antibodies for AMDHD2 protein, ACACB protein, DGKH protein, DDX20 protein and CDK11B protein, which can be used to specifically detect the expression level of the corresponding protein.
[0064] In one embodiment of the present disclosure, the long new crown typing product includes a reagent for detecting the level of mRNA of each protein in the biomarker group. In one example, the biomarker group can be mRNA of AMDHD2 protein, mRNA of ACACB protein, mRNA of DGKH protein, mRNA of DDX20 protein, and mRNA of CDK11B protein. In another example, the biomarker group can also be mRNA of FPGT protein, mRNA of CBX6 protein, mRNA of DGKH protein, mRNA of RBBP4 protein, and mRNA of CRH protein.
[0065] Further, the long new crown typing product includes but is not limited to a kit, a detection reagent or a biochip. Exemplary, the reagent used for PCR detection of the levels of AMDHD2 protein mRNA, ACACB protein mRNA, DGKH protein mRNA, DDX20 protein mRNA and CDK11B protein mRNA. Further, the PCR detection can use reverse transcription quantitative PCR (qRT-PCR).
[0066] The present disclosure also provides a method for typing long-term COVID-19, which comprises the following steps:
[0067] S1. Obtain a sample to be tested, wherein the sample to be tested is taken from a patient with COVID-19;
[0068] S2. Using the long new crown typing product described in claim 7 or 8, detecting the protein expression levels of the first biomarker, the second biomarker, the third biomarker, the fourth biomarker, and the fifth biomarker in the sample to be tested, and obtaining the protein expression results of the sample to be tested;
[0069] wherein the first biomarker is selected from at least one biomarker in the first biomarker subgroup; the second biomarker is selected from at least one biomarker in the second biomarker subgroup; the third biomarker is selected from at least one biomarker in the third biomarker subgroup; the fourth biomarker is selected from at least one biomarker in the fourth biomarker subgroup; and the fifth biomarker is selected from at least one biomarker in the fifth biomarker subgroup;
[0070] S3, inputting the protein expression results of the sample to be tested obtained in step S2 into the long-term new crown multi-neural network language diagnosis model to obtain the output results of the long-term new crown multi-neural network language diagnosis model;
[0071] If the output result is the first biomarker, the sample to be tested is of the nervous system type; if the output result is the second biomarker, the sample to be tested is of the composite type of the cardiovascular system and the nervous system; if the output result is the third biomarker, the sample to be tested is of the composite type of the skeletal muscle system and the whole body nonspecific system; if the output result is the fourth biomarker, the sample to be tested is of the composite type of the cardiovascular system and the respiratory system; if the output result is the fifth biomarker, the sample to be tested is of the multi-system composite type, and the multi-system composite type includes the composite type of at least four systems including the nervous system, the circulatory system, the respiratory system, the skeletal muscle system, and the whole body nonspecific system.
[0072] In one embodiment of the present disclosure, in step S1, the sample to be tested obtained may be a serum sample of peripheral blood.
[0073] In step S2, the above-mentioned long new crown typing product is used to detect the expression levels of AMDHD2 protein, ACACB protein, DGKH protein, DDX20 protein and CDK11B protein in the sample to be tested to obtain the protein expression results of the sample to be tested.
[0074] In step S3, the protein expression results of the sample to be tested obtained in step S2 are input into the long new crown multivariate neural network language diagnosis model to obtain the output result of the long new crown multivariate neural network language diagnosis model. If the output result is AMDHD2 protein, the sample to be tested is a nervous system type. If the output result is ACACB protein, the sample to be tested is a composite type of cardiovascular system and nervous system. If the output result is DDX20 protein, the sample to be tested is a composite type of cardiovascular system and respiratory system. If the output result is DGKH protein, the sample to be tested is a composite type of skeletal muscle system and systemic nonspecific system. If the output result is CDK11B protein, the sample to be tested is a multi-system composite type.
[0075] In another embodiment of the present disclosure, in step S1, the sample to be tested may be peripheral blood.
[0076] In step S2, the long new crown typing product described in claim 7 or 8 is used to detect the expression levels of FPGT protein, CBX6 protein, DGKH protein, RBBP4 protein and CRH protein in the sample to be tested to obtain the protein expression results of the sample to be tested.
[0077] In step S3, the protein expression results of the sample to be tested obtained in step S2 are input into the long new crown multivariate neural network language diagnosis model to obtain the output result of the long new crown multivariate neural network language diagnosis model; if the output result is FPGT protein, the sample to be tested is a nervous system type; if the output result is CBX6 protein, the sample to be tested is a composite type of cardiovascular system and nervous system; if the output result is RBBP4 protein, the sample to be tested is a composite type of cardiovascular system and respiratory system; if the output result is DGKH protein, the sample to be tested is a composite type of skeletal muscle system and systemic nonspecific system; if the output result is CRH protein, the sample to be tested is a multi-system composite type.
[0078] In one embodiment of the present disclosure, the construction process of the long-term new crown multi-neural network language diagnosis model includes:
[0079] Step 10, obtain samples from the subjects with long new crown, and divide the subjects with long new crown into five experimental groups. The experimental groups include: cardiovascular system and respiratory system composite long new crown group (CAPM group), cardiovascular system and nervous system composite long new crown group (CACRB group), nervous system type long new crown group (NEU group), skeletal muscle system and whole body nonspecific system composite long new crown group (MSK+SYST group), multi-system composite long new crown group (MULTI group).
[0080] Step 20, obtaining the protein expression levels of the first biomarker, the second biomarker, the third biomarker, the fourth biomarker and the fifth biomarker of each sample in each experimental group; the protein expression levels of the first biomarker, the second biomarker, the third biomarker, the fourth biomarker and the fifth biomarker of each sample in each experimental group constitute a data set.
[0081] Step 30: Randomly extract 70% to 80% of the data in the data set as the training set, input the training set into the multivariate neural network language diagnosis model for iterative training, and obtain the long-term new crown multivariate neural network language diagnosis model.
[0082] For example, in step 20, the expression levels of AMDHD2 protein, ACACB protein, DGKH protein, DDX20 protein and CDK11B protein of each sample in each experimental group are obtained. The expression levels of AMDHD2 protein, ACACB protein, DGKH protein, DDX20 protein and CDK11B protein of each sample in each experimental group constitute the first data set. It should be noted that each group corresponds to the expression levels of AMDHD2 protein, ACACB protein, DGKH protein, DDX20 protein and CDK11B protein of the sample in the group. For example, the expression levels of AMDHD2 protein, ACACB protein, DGKH protein, DDX20 protein and CDK11B protein corresponding to each sample in the CAPM group constitute the first data combination. Similarly, the expression levels of AMDHD2 protein, ACACB protein, DGKH protein, DDX20 protein and CDK11B protein corresponding to each sample of the CACRB group constitute the second data combination; the expression levels of AMDHD2 protein, ACACB protein, DGKH protein, DDX20 protein and CDK11B protein corresponding to each sample of the NEU group constitute the third data combination; the expression levels of AMDHD2 protein, ACACB protein, DGKH protein, DDX20 protein and CDK11B protein corresponding to each sample of the MSK+SYST group constitute the fourth data combination; the expression levels of AMDHD2 protein, ACACB protein, DGKH protein, DDX20 protein and CDK11B protein corresponding to each sample of the MULTI group constitute the fifth data combination. The first data combination to the fifth data combination constitute the first data set.
[0083] In step 30, 70% to 80% of the data in the first data set are randomly selected as the first training set, and another 20% to 30% of the data are used as the first verification set. For example, 70% of the data can be extracted from the first data combination to the fifth data combination as the first training set, and the remaining 30% of the data can be used as the first verification set. The first training set is input into the multivariate neural network language diagnosis model for iterative training to obtain the long new crown multivariate neural network language diagnosis model. The first verification set can be used to verify the accuracy and specificity of the long new crown multivariate neural network language diagnosis model. The first verification set verifies that the accuracy and specificity of the long new crown multivariate neural network language diagnosis model meet the preset standards, and the long new crown multivariate neural network language diagnosis model is successfully constructed.
[0084] For another example, in step 20, the expression levels of FPGT protein, CBX6 protein, DGKH protein, RBBP4 protein and CRH protein of each sample in each experimental group are obtained. The expression levels of FPGT protein, CBX6 protein, DGKH protein, RBBP4 protein and CRH protein of each sample in each experimental group constitute the second data set. It should be noted that each group corresponds to the expression levels of FPGT protein, CBX6 protein, DGKH protein, RBBP4 protein and CRH protein of the sample in the group. For example, the expression levels of FPGT protein, CBX6 protein, DGKH protein, RBBP4 protein and CRH protein corresponding to each sample of the CAPM group constitute the sixth data combination. Similarly, the expression levels of FPGT protein, CBX6 protein, DGKH protein, RBBP4 protein and CRH protein corresponding to each sample of the CACRB group constitute the seventh data combination; the expression levels of FPGT protein, CBX6 protein, DGKH protein, RBBP4 protein and CRH protein corresponding to each sample of the NEU group constitute the eighth data combination; the expression levels of FPGT protein, CBX6 protein, DGKH protein, RBBP4 protein and CRH protein corresponding to each sample of the MSK+SYST group constitute the ninth data combination; the expression levels of FPGT protein, CBX6 protein, DGKH protein, RBBP4 protein and CRH protein corresponding to each sample of the MULTI group constitute the tenth data combination. The sixth data combination to the tenth data combination constitute the second data set.
[0085] In step 30, 70% to 80% of the data in the second data set is randomly selected as the second training set, and another 20% to 30% of the data is used as the second verification set. For example, 80% of the data can be extracted from the first data combination to the fifth data combination as the second training set, and the remaining 20% of the data can be used as the second verification set. The second training set is input into the multivariate neural network language diagnosis model for iterative training to obtain the long new crown multivariate neural network language diagnosis model. The second verification set can be used to verify the accuracy and specificity of the long new crown multivariate neural network language diagnosis model. The second verification set verifies that the accuracy and specificity of the long new crown multivariate neural network language diagnosis model meet the preset standards, and the long new crown multivariate neural network language diagnosis model is successfully constructed.
[0086] The long-term new coronary disease typing biomarker group provided by the present invention is further explained below in conjunction with specific embodiments.
[0087] 1. Experimental samples and experimental groups
[0088] Peripheral blood samples were collected from the subjects. Among them, the basis for the diagnosis of long-term COVID-19 in the subjects was: persistent or new symptoms appeared 3 months after COVID-19 infection, including fatigue, shortness of breath and cognitive dysfunction, which lasted for at least 2 months and could not be explained by other reasons.
[0089] The subjects were classified as follows: subjects who showed symptoms of both the cardiovascular and respiratory systems were classified as the cardiovascular and respiratory complex long new crown group (CAPM group); subjects who showed symptoms of both the cardiovascular and nervous systems were classified as the cardiovascular and nervous system complex long new crown group (CACRB group); subjects who mainly showed symptoms of the nervous system were classified as the nervous system type long new crown group (NEU group); subjects who showed symptoms of the skeletal muscle system and systemic nonspecific system were classified as the skeletal muscle system and systemic nonspecific system complex long new crown group (MSK+SYST group); subjects who showed symptoms of four or more systems were classified as the multi-system complex long new crown group (MULTI group); subjects who had no long new crown symptoms were classified as the non-long new crown group (NLC group). Peripheral blood samples were collected from each subject and could be used directly in the experiment or stored at -80°C for future use.
[0090] The classification of each subject is mainly based on clinical symptoms. For example, throat discomfort, nasal congestion, cough and sputum, and dyspnea are classified as respiratory system type; chest tightness and palpitations are classified as cardiovascular system type; fatigue, sleep disorders, dizziness, impaired vision, headache, taste disorders, olfactory disorders, and cognitive disorders are classified as nervous system type; myalgia and joint pain are classified as skeletal muscle system type; hair loss and rash are classified as systemic nonspecific system type; loss of appetite is classified as a digestive system symptom.
[0091] In this way, the collected subject samples were classified, including 30 cases in the CAPM group, 37 cases in the CACRB group, 56 cases in the NEU group, 35 cases in the MSK+SYST group, 43 cases in the MULTI group, and 56 cases in the NLC group in the present disclosed embodiment.
[0092] 2. Transcriptome Analysis
[0093] Step 1, RNA isolation: Extract total RNA from peripheral blood samples in the CAPM group, CACRB group, NEU group, MSK+SYST group, MULTI group, and NLC group. Total RNA was extracted using the method provided in the instruction manual of TRIzol reagent (Invitrogen, California, USA).
[0094] Step 2: Preparation of RNA library: Based on the total RNA of each sample, the VAHTS Universal V6 RNA-seq library preparation kit was used to construct the RNA library corresponding to each sample.
[0095] Step 3, RNA sequencing: The RNA libraries corresponding to each sample in the CAPM group, CACRB group, NEU group, MSK+SYST group, MULTI group and NLC group were sequenced on the Illumina Novaseq 6000 platform, and each sample generated approximately 150bp of paired-end reads, which were the original sequencing data. The fastp software was used to process the original sequencing data in fastq format using the default parameters, and HISAT was used for sequence alignment after removing low-quality reads. The reference genome was GRCh38.p13. Finally, the expression level of each gene in each sample was counted and quantified by HTSeq-count in FPKM to obtain the gene expression level of each sample.
[0096] Step 4, differential gene expression analysis: The DESeq2 R software package was used to perform differential expression analysis on the gene expression levels of each sample obtained by RNA sequencing based on default parameters. The p value <0.05 and foldchange>2 or foldchange<-2 were used as the thresholds for significantly differentially expressed genes (DEGs). For example, according to the thresholds for significantly differentially expressed genes (DEGs) set above, the significantly differentially expressed genes obtained by comparing the CAPM group with the CACRB group, the NEU group, the MSK+SYST group, the MULTI group, and the NLC group were the significantly differentially expressed genes (DEGs) of the CAPM group; for another example, the significantly differentially expressed genes obtained by comparing the MULTI group with the CAPM group, the CACRB group, the NEU group, the MSK+SYST group, and the NLC group were the significantly differentially expressed genes (DEGs) of the MULTI group. In other words, the significantly differentially expressed genes obtained by comparing a group with other groups other than the group are the significantly differentially expressed genes (DEGs) of the group. Thus, the significantly differentially expressed genes (DEGs) of each group were obtained through differential gene expression analysis.
[0097] 3. Proteomic Analysis
[0098] Step S10, plasma protein extraction: the collected peripheral blood samples were centrifuged to obtain plasma samples of each subject in the CAPM group, CACRB group, NEU group, MSK+SYST group, MULTI group and NLC group, and plasma proteins were extracted.
[0099] Step S20, enzymatic hydrolysis: a portion of the plasma protein obtained in step S10 is enzymatically hydrolyzed (for example, using trypsin for enzymatic hydrolysis) to obtain a mixed first polypeptide solution.
[0100] Step S30, enriching low-abundance plasma proteins: The other part of the plasma proteins obtained in step S10 was enriched with the EasyPeptide DeepP low-abundance protein enrichment and pretreatment preparation kit (OSFP0002) (Omicsolution, Shanghai, China) to obtain low-abundance proteins in the plasma proteins. The enriched proteins were reduced and alkylated, enzymatically hydrolyzed into peptides, and desalted to obtain a second polypeptide solution.
[0101] It is understandable that the order of step S20 and step S30 can be swapped, and the second polypeptide solution can be obtained first, and then the first polypeptide solution.
[0102] Step S40, mass spectrometry analysis: The first polypeptide solution corresponding to the same sample obtained in step S20 and the second polypeptide solution corresponding to the same sample obtained in step S30 are mixed to obtain polypeptide solutions of each sample. Liquid chromatography-mass spectrometry is used to perform mass spectrometry analysis on the polypeptide segments in the polypeptide solution, and the signal intensities of the polypeptide segments in different samples are compared to relatively quantify the proteins corresponding to the polypeptide segments. The abundance (amount) of each polypeptide segment is proportional to the peak area or signal intensity of the polypeptide segment in the mass spectrum; and the retention time (RT) of the same polypeptide segment in different samples on liquid chromatography is consistent. Therefore, by comparing the information in the spectral library and integrating it into the corresponding protein, a quantitative analysis of the relative expression level of the protein in different samples can be achieved. In this way, the protein expression level of each sample in each group is obtained.
[0103] Step S50, differential protein analysis: The protein expression level of each sample in each group is analyzed using the limma R software package. Trend parameters are selected to improve the stability of the analysis, and FDR<0.05 and Foldchange>1.5, or other thresholds are used as the threshold of differential protein (with significant differences). For example, according to the threshold of the above-mentioned differential protein, the significant difference protein obtained by comparing the CACRB group with the CAPM group, the NEU group, the MSK+SYST group, the MULTI group, and the NLC group is the differential protein of the CACRB group; for another example, the significant difference protein obtained by comparing the NEU group with the CAPM group, the CACRB group, the MULTI group, the MSK+SYST group, and the NLC group is the differential protein of the NEU group. In other words, a group is compared with other groups other than the group, and the significant difference protein obtained is the differential protein of the group. In this way, the differential proteins of each group are obtained by differential protein analysis.
[0104] 4. Genomic and proteomic data analysis
[0105] (1) Data integration analysis based on similarity network fusion
[0106] Similarity Network Fusion (SNF) was used to perform multi-omics data joint analysis on the proteomics and transcriptomics of the CACRB group, CAPM group, NEU group, MSK+SYST group, MULTI group, and NLC group. Figure 1 The sample-by-sample similarity matrix for the subjects shown, Figure 1 0 to 4 indicates positive correlation, 0 to -4 indicates negative correlation, and SNF1 to SNF 6 indicate the six groups (Muti-omics cluster) obtained by clustering samples using the similarity network fusion algorithm. Figure 2for Figure 1 A partial enlarged schematic diagram of Figure 2 Each small square in the box corresponds to a sample. The results showed that the protein expression level characteristics of the subjects had a high positive correlation with the clinical-based grouping, that is, the proteomic characteristics of the subjects were highly consistent with the clinical subtype classification defined by each group, suggesting that there are protein markers that can be used as a basis for clinical classification.
[0107] (2) Joint proteome-transcriptome analysis
[0108] Step 100, using the GSEA function in the cluster Profiler R software package to perform gene function enrichment analysis and protein function enrichment analysis on the highly expressed significantly differentially expressed genes (DEGs) and differentially expressed proteins in each group, respectively. The enrichment analysis uses the Reactome pathway database. In each group, screen for enriched signal pathways whose protein function enrichment analysis score (score represented by NES) is greater than 0 and whose gene function enrichment analysis score (score represented by NES) is greater than 0. Figure 3 As shown in the figure, the enriched signal pathways obtained in each group and the enriched protein types contained in the pathways were integrated into the signal pathways, and a total of 28 related enriched signal pathways were obtained. In other words, the genes corresponding to the enriched proteins of the 28 signal pathways are also significantly differentially expressed genes (DEGs).
[0109] Step 200, obtain the enriched proteins corresponding to 28 signal pathways, and compare the expression level differences and ROC curves of different enriched proteins between groups according to the protein expression levels of each sample measured in the proteome analysis step S40, and set the screening criteria as follows: the statistical difference in the expression level of the enriched proteins between the groups satisfies P < 0.001 and the area under the ROC curve AUC > 0.85, such as Figures 4 to 8 As shown, significant marker proteins of each experimental group (experimental groups include CAPM group, CACRB group, NEU group, MSK+SYST group, and MULTI group) were screened out.
[0110] The results of the screening of significant marker proteins are as follows: in the MULTI group, proteins related to stress response (CRH protein), cell cycle regulation (CDK11B protein), and ether lipid metabolism (LPCAT2 protein) were upregulated; Figure 4 As shown in Figure 2, the significant marker proteins screened by the MULTI group can include CDK11B protein, CRH protein and LPCAT2 protein. In the NEU group, proteins related to sugar metabolism and glycoprotein regulation were increased, such as FPGT protein and AMDHD2 protein; Figure 5As shown in Figure 2, the significant marker proteins screened out in the NEU group can include FPGT protein and AMDHD2 protein. In the MSK+SYST group, the levels of proteins such as DGKH protein and HNRNPUL1 protein that promote glycerophospholipid metabolism were increased; Figure 6 As shown, the significant marker proteins screened out by the MSK+SYST group can include HNRNPUL1 protein and DGKH protein. In the CACRB group, the levels of proteins related to macrophage inhibition were increased, such as TSC2 protein, CBX6 protein, and fatty acid synthesis protein ACACB protein; Figure 7 As shown in Figure 2, the significant marker proteins screened out by the CACRB group can include PSME4 protein, TSC2 protein, CBX6 protein and ACACB protein. The level of DDX2 protein involved in downregulating the NF-κB signaling pathway was increased in the CAPM group; Figure 8 As shown, the significant marker proteins screened by the CAPM group may include DDX20 protein, RBBP4 protein and PCNA protein. Among them, Figure 4-Figure 8 The protein quantitation (Protein quantitation) in the vertical axis is the relative protein abundance (log2).
[0111] Step 300: Randomly select a protein from the significant marker proteins of each experimental group to obtain a significant marker protein group containing five proteins. In other words, multiple significant marker protein groups can be obtained through permutations and combinations. Screen each significant marker protein group to select the significant marker protein group with the best long-term new crown typing accuracy and specificity.
[0112] Exemplarily, taking the significant landmark protein group including CDL11B protein, FPGT protein, HNRNPUL1 protein, PSME4 protein and DDX20 protein as an example, the screening process includes: according to the protein expression level in each sample in each group measured in step S40 in the proteome analysis, the expression levels of CDL11B protein, FPGT protein, HNRNPUL1 protein, PSME4 protein and DDX20 protein corresponding to each sample in each experimental group (CACRB group, CAPM group, NEU group, MSK+SYST group, MULTI group) are used as data sets. 70% are randomly selected from the data set as a training set, and the remaining 30% are used as a test set. The training set is input into the multivariate neural network language diagnosis model, the multivariate neural network language diagnosis model is trained, and the long new crown multivariate neural network language diagnosis model is obtained. The test set was used to verify the long-term new crown multi-neural network language diagnosis model, and the accuracy and specificity of the long-term new crown multi-neural network language diagnosis model were judged according to the area under the ROC curve corresponding to the measured CDL11B protein, FPGT protein, HNRNPUL1 protein, PSME4 protein and DDX20 protein.
[0113] Result analysis: According to the validation results of each significant marker protein group, such as Fig. 9 As shown, the MULTI group selected CDK11B protein, the NEU group selected AMDHD2 protein, the CACRB group selected ACACB protein, the MSK+SYST group selected DGKH protein, and the CAPM group selected DDX20 protein. The long-term new crown multivariate neural network language diagnosis model constructed through continuous iterative training has high accuracy and specificity. Fig. 9 As shown in the figure, according to the area under the curve (AUC) of the receiver operating characteristic (ROC) curve, the AUC value of CDK11B protein in the MULTI group was 0.96; the AUC value of AMDHD2 protein in the NEU group was 0.84; the AUC value of ACACB protein in the CACRB group was 0.95; the AUC value of DGKH protein in the MSK+SYST group was 0.92; and the AUC value of DDX20 protein in the CAPM group was 0.91. Fig.10 As shown, the expression level of CDK11B protein in the MULTI group was significantly higher than that in other groups; the expression level of AMDHD2 protein in the NEU group was significantly higher than that in other groups; the expression level of ACACB protein in the CACRB group was significantly higher than that in other groups; the expression level of DGKH protein in the MSK+SYST group was significantly higher than that in other groups; and the expression level of DDX20 protein in the CAPM group was significantly higher than that in other groups. In this way, CDK11B protein, AMDHD2 protein, ACACB protein, DGKH protein and DDX20 protein were combined to form a biomarker group, which further verified the accuracy of the selection of protein markers for long-term new crown typing.
[0114] like Fig.11 As shown in the figure, the MULTI group selected CRH protein, the NEU group selected FPGT protein, the CACRB group selected CBX6 protein, the MSK+SYST group selected DGKH protein, and the CAPM group selected RBBP4 protein. The long new crown multivariate neural network language diagnosis model constructed through continuous iterative training also has high accuracy and specificity. According to the area under the curve (AUC) of the receiver operating characteristic curve (ROC), the CRH protein in the MULTI group had an AUC value of 0.92; the FPGT protein in the NEU group had an AUC value of 0.86; the CBX6 protein in the CACRB group had an AUC value of 0.91; the DGKH protein in the MSK+SYST group had an AUC value of 0.93; and the RBBP4 protein in the CAPM group had an AUC value of 0.91.
[0115] It is understandable that the above exemplary MULTI group selects CDK11B protein, NEU group selects AMDHD2 protein, CACRB group selects ACACB protein, MSK+SYST group selects DGKH protein, and CAPM group selects DDX20 protein to form a biomarker group; or, the MULTI group selects CRH protein, the NEU group selects FPGT protein, the CACRB group selects CBX6 protein, the MSK+SYST group selects DGKH protein, and the CAPM group selects RBBP4 protein to form another biomarker group. The biomarker group provided in the present disclosure is not limited to the above two biomarker groups, and two or more biomarkers can also be selected in a long new crown typing (such as MULTI, NEU, CACRB, MSK+SYST or CAPM) to form a biomarker group for typing long new crown.
[0116] Those skilled in the art will readily appreciate other embodiments of the present disclosure after considering the specification and practicing the invention disclosed herein. This application is intended to cover any modification, use or adaptation of the present disclosure, which follows the general principles of the present disclosure and includes common knowledge or customary techniques in the art that are not disclosed in the present disclosure. The specification and examples are intended to be exemplary only, and the true scope and spirit of the present disclosure are indicated by the appended claims.
Claims
1. A biomarker panel for long-term COVID-19 typing, characterized in that: include: at least one of the biomarkers FPGT, AMDHD2 of the first biomarker subgroup; at least one of the biomarkers PSME4, TSC2, CBX6, ACACB of the second biomarker subgroup; at least one of the biomarkers HNRNPUL1 and DGKH of the third biomarker subgroup; at least one of the biomarkers DDX20, RBBP4, and PCNA of the fourth biomarker subgroup; and, at least one of the biomarkers CDK11B, CRH, LPCAT2 of the fifth biomarker subgroup.
2. The biomarker panel for long-term COVID-19 typing according to claim 1, characterized in that: The biomarker panel includes AMDHD2 protein, ACACB protein, DGKH protein, DDX20 protein and CDK11B protein.
3. The biomarker panel for long-term COVID-19 typing according to claim 1, characterized in that: The biomarker panel includes mRNA of AMDHD2 protein, mRNA of ACACB protein, mRNA of DGKH protein, mRNA of DDX20 protein and mRNA of CDK11B protein.
4. The biomarker panel for long-term COVID-19 typing according to claim 1, characterized in that: The biomarker panel includes FPGT protein, CBX6 protein, DGKH protein, RBBP4 protein and CRH protein.
5. The biomarker panel for long-term COVID-19 typing according to claim 1, characterized in that: The biomarker panel includes mRNA of FPGT protein, mRNA of CBX6 protein, mRNA of DGKH protein, mRNA of RBBP4 protein and mRNA of CRH protein.
6. Use of the biomarker panel described in any one of claims 1-5 in the preparation of long-term new coronary typing products.
7. A long new crown typing product, characterized in that, Comprising reagents for detecting each biomarker in the biomarker panel according to any one of claims 1 to 5.
8. The long new crown typing product according to claim 7, characterized in that The product is a test kit, a detection reagent or a biochip.
9. A method for typing long coronavirus, characterized in that: The following steps are involved: S1. Obtain a sample to be tested, wherein the sample to be tested is taken from a patient with COVID-19; S2. Using the long new crown typing product described in claim 7 or 8, detecting the protein expression levels of the first biomarker, the second biomarker, the third biomarker, the fourth biomarker, and the fifth biomarker in the sample to be tested, and obtaining the protein expression results of the sample to be tested; wherein the first biomarker is selected from at least one biomarker in the first biomarker subgroup; the second biomarker is selected from at least one biomarker in the second biomarker subgroup; the third biomarker is selected from at least one biomarker in the third biomarker subgroup; the fourth biomarker is selected from at least one biomarker in the fourth biomarker subgroup; and the fifth biomarker is selected from at least one biomarker in the fifth biomarker subgroup; S3, inputting the protein expression results of the sample to be tested obtained in step S2 into the long-term new crown multi-neural network language diagnosis model to obtain the output results of the long-term new crown multi-neural network language diagnosis model; If the output result is the first biomarker, the sample to be tested is of the nervous system type; if the output result is the second biomarker, the sample to be tested is of the composite type of the cardiovascular system and the nervous system; if the output result is the third biomarker, the sample to be tested is of the composite type of the skeletal muscle system and the whole body nonspecific system; if the output result is the fourth biomarker, the sample to be tested is of the composite type of the cardiovascular system and the respiratory system; if the output result is the fifth biomarker, the sample to be tested is of the multi-system composite type, and the multi-system composite type includes the composite type of at least four systems including the nervous system, the circulatory system, the respiratory system, the skeletal muscle system, and the whole body nonspecific system.
10. The long COVID-19 typing method according to claim 9, characterized in that: The following steps are involved: S1. Obtain a sample to be tested, wherein the sample to be tested is taken from a patient with COVID-19; S2. Use the long new crown typing product described in claim 7 or 8 to detect the expression levels of AMDHD2 protein, ACACB protein, DGKH protein, DDX20 protein and CDK11B protein in the sample to be tested to obtain the protein expression results of the sample to be tested; S3. Input the protein expression results of the sample to be tested obtained in step S2 into the long-term new crown multivariate neural network language diagnosis model to obtain the output result of the long-term new crown multivariate neural network language diagnosis model; if the output result is AMDHD2 protein, the sample to be tested is a nervous system type; if the output result is ACACB protein, the sample to be tested is a composite type of cardiovascular system and nervous system; if the output result is DDX20 protein, the sample to be tested is a composite type of cardiovascular system and respiratory system; if the output result is DGKH protein, the sample to be tested is a composite type of skeletal muscle system and systemic nonspecific system; if the output result is CDK11B protein, the sample to be tested is a multi-system composite type.
11. The long COVID-19 typing method according to claim 9, characterized in that: The following steps are involved: S1. Obtain a sample to be tested, wherein the sample to be tested is taken from a patient with COVID-19; S2. Use the long new crown typing product described in claim 7 or 8 to detect the expression levels of FPGT protein, CBX6 protein, DGKH protein, RBBP4 protein and CRH protein in the sample to be tested to obtain the protein expression results of the sample to be tested; S3, inputting the protein expression result of the sample to be tested obtained in step S2 into the long new crown multi-neural network language diagnosis model to obtain the output result of the long new crown multi-neural network language diagnosis model; if the output result is FPGT protein, the sample to be tested is a nervous system type; if the output result is CBX6 protein, the sample to be tested is a composite type of cardiovascular system and nervous system; If the output result is RBBP4 protein, the sample to be tested is a composite type of cardiovascular system and respiratory system; if the output result is DGKH protein, the sample to be tested is a composite type of skeletal muscle system and systemic nonspecific system; if the output result is CRH protein, the sample to be tested is a composite type of multiple systems.
12. The long COVID-19 typing method according to claim 9, characterized in that: The construction process of the long-term new crown multi-neural network language diagnosis model in step S3 is: Samples of long-COVID subjects were obtained, and the long-COVID subjects were divided into five experimental groups, including: a cardiovascular system and respiratory system composite long-COVID group, a cardiovascular system and nervous system composite long-COVID group, a nervous system type long-COVID group, a skeletal muscle system and systemic nonspecific system composite long-COVID group, and a multi-system composite long-COVID group; Obtaining the protein expression levels of the first biomarker, the second biomarker, the third biomarker, the fourth biomarker and the fifth biomarker of each sample in each experimental group; the protein expression levels of the first biomarker, the second biomarker, the third biomarker, the fourth biomarker and the fifth biomarker of each sample in each experimental group constitute a data set; 70% to 80% of the data in the data set were randomly selected as the training set, and the training set was input into the multivariate neural network language diagnosis model for iterative training to obtain the long-term new crown multivariate neural network language diagnosis model.
