Prediction marker and prediction method for acute upper respiratory infection susceptibility

Abstract

The invention discloses a prediction marker and a prediction method for acute upper respiratory infection susceptibility. The prediction marker and the prediction method are used for solving the problem that current acute upper respiratory infection susceptibility risk prediction is low in accuracy. The method comprises the following steps: firstly, providing a prediction marker for acute upper respiratory infection susceptibility, wherein the prediction marker is a prediction feature combination obtained by comprehensively considering human health state blood routine data and baseline blood transcriptome data and carrying out feature processing; secondly, providing a prediction method, and obtaining a to-be-detected blood sample; and performing blood analysis based on a to-be-detected blood sample, inputting the obtained blood routine sample and the gene expression sample into a susceptibility prediction model pre-trained based on a prediction marker for susceptibility risk judgment, and obtaining a susceptibility risk assessment result of acute upper respiratory infection of a human body corresponding to the to-be-detected blood sample in a future period of time, therefore, the acute upper respiratory infection susceptibility risk of the human body in a period of time in the future can be simply, quickly and accurately predicted.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of upper respiratory tract infection susceptibility research, and particularly relates to a prediction marker for acute upper respiratory tract infection susceptibility, a prediction method for acute upper respiratory tract infection susceptibility, a prediction device for acute upper respiratory tract infection susceptibility, an electronic device and a storage medium. BACKGROUND

[0002] Acute upper respiratory tract infection (commonly referred to as upper respiratory tract infection) is a general term for a group of acute inflammatory diseases of the nasal cavity, pharynx and larynx. The clinical manifestations are cough, sore throat, nasal congestion, runny nose and other symptoms. About 80% of acute upper respiratory tract infections are caused by viruses, including rhinovirus, coronavirus, influenza virus, parainfluenza virus and respiratory syncytial virus. Another 20% are caused by bacteria. Adults have an average of 2-4 upper respiratory tract infections per year. Acute upper respiratory tract infection is usually self-limiting and has a good prognosis. However, it has a high incidence, affects work and life, and can sometimes be accompanied by serious complications and has certain infectivity.

[0003] From an individual perspective, the susceptibility of upper respiratory tract infection directly determines the risk of individuals suffering from such infections, the course of the disease and possible complications (such as severe lower respiratory tract infection, pneumonia, acute exacerbation of asthma or cardiovascular events). Identifying such high-susceptibility individuals helps to give more precise interventions in health status. For example, vaccination, regular health monitoring, lifestyle guidance and individualized drug prevention strategies, thereby reducing the disease burden and improving prognosis. From a population perspective, the differences in upper respiratory tract infection susceptibility among different individuals in the population are of great significance to the allocation of resources of the entire public health system, the construction of epidemiological prediction models and the formulation of disease prevention and control policies. In the population, if there is a certain proportion of high-susceptibility individuals, during the epidemic period of the pathogen, this group of people may become a "virus amplification device", accelerating the spread of the disease and increasing the overall infection burden of society. Therefore, carrying out susceptibility identification and stratified management (such as prioritizing high-risk groups for vaccination, health education and post-exposure prevention) helps to improve the efficiency of prevention and control, slow down the spread of the disease and protect the medical system from overload.

[0004] Current studies on susceptibility to upper respiratory tract infection are mainly conducted through virus challenge experiments. By inoculating healthy volunteers with rhinovirus, respiratory syncytial virus, coronavirus or influenza virus, the infection and illness of the individuals 4-10 days after inoculation are observed to explain the short-term susceptibility mechanism of specific pathogens and predict susceptibility. However, such studies are limited by highly controlled experimental conditions and a short observation window of 4-10 days, and are difficult to reflect the cumulative effects of dynamic multi-pathogen exposure (different pathogens are infected each time a cold is caught, and it is not a single pathogen), environmental factors (changes in environmental temperature, humidity and air pollutant concentration), and individual long-term social and psychological factors (individual lifestyle, stress, sleep, diet, etc.) in the real environment, resulting in low prediction accuracy. SUMMARY

[0005] The present application provides a prediction marker for susceptibility to acute upper respiratory tract infection, a prediction method for susceptibility to acute upper respiratory tract infection, a prediction device for susceptibility to acute upper respiratory tract infection, an electronic device and a storage medium, which are used to solve or partially solve the technical problem of low prediction accuracy of current susceptibility risk to acute upper respiratory tract infection.

[0006] The present application provides a prediction marker for susceptibility to acute upper respiratory tract infection, which is a prediction feature combination obtained by comprehensively considering the health status blood routine data and baseline blood transcriptome data and performing feature processing.

[0007] Optionally, the prediction feature combination includes a blood routine index group and a differential expression gene group; the blood routine index group is obtained by statistical difference analysis based on the health status blood routine data; and the differential expression gene group is obtained by gene differential expression analysis and feature screening based on the baseline blood transcriptome data.

[0008] Optionally, the blood routine index group includes RDW-CV, P-LCR, MONO, Hb and MPV; and the differential expression gene group includes A2M, TBC1D30, MCIR, PHOSPHO1, TLCD4, B3GALNT1, TACSTD2 and ZNF154.

[0009] The present application also provides a prediction method for susceptibility to acute upper respiratory tract infection, which uses the prediction marker as described above to predict the susceptibility risk of human body to acute upper respiratory tract infection; and the method comprises the following steps:

[0010] Obtaining a blood sample to be tested;

[0011] Performing blood analysis based on the blood sample to be tested to obtain a blood routine sample and a gene expression sample;

[0012] inputting the blood routine sample and the gene expression sample into a susceptibility prediction model pre-trained based on the prediction marker to determine the susceptibility risk, and obtaining a susceptibility risk evaluation result of the test blood sample corresponding to the human body in a future period of time.

[0013] Optionally, the process of constructing and training the susceptibility prediction model comprises:

[0014] obtaining a pre-constructed blood routine dataset and a baseline transcriptome dataset; the blood routine dataset corresponds to a group of blood routine indexes with statistical differences; and the baseline transcriptome dataset corresponds to a group of differential expression genes with gene expression differences;

[0015] dividing the blood routine dataset into a blood routine training set and a blood routine test set according to a preset division ratio, and dividing the baseline transcriptome dataset into a baseline transcriptome training set and a baseline transcriptome test set;

[0016] merging the group of blood routine indexes and the group of differential expression genes to construct a fusion feature matrix;

[0017] combining the fusion feature matrix as a prediction feature group of a susceptibility risk prediction model, taking acute upper respiratory tract infection susceptibility as a target variable, and constructing an initial susceptibility prediction model;

[0018] training the initial susceptibility prediction model based on the blood routine training set and the baseline transcriptome training set to determine the susceptibility risk, and adjusting and optimizing the model parameters during the training process;

[0019] evaluating the model prediction performance of the susceptibility prediction model obtained after the training by using the blood routine test set and the baseline transcriptome test set.

[0020] Optionally, the process of constructing the baseline transcriptome dataset comprises:

[0021] obtaining baseline blood transcriptome data of a first test population and first symptom condition data of the first test population related to acute upper respiratory tract infection in a historical period;

[0022] dividing the first test population into a first susceptible group and a first non-susceptible group according to the first symptom condition data;

[0023] preprocessing the baseline blood transcriptome data to obtain a gene expression matrix and a plurality of PBMC samples;

[0024] Screening protein-coding genes with expression ratios greater than or equal to a preset expression ratio from the gene expression matrix, and normalizing the protein-coding genes by TPM standardization to obtain a target gene expression matrix;

[0025] Performing gene differential expression analysis on the first susceptible group and the first non-susceptible group based on the target gene expression matrix, and screening a plurality of differentially expressed genes meeting a preset expression change condition through feature screening to form a differentially expressed gene group affecting the susceptibility risk of acute upper respiratory tract infection in humans;

[0026] Extracting gene expression data of the differentially expressed gene group in each PBMC sample from the gene expression profile, and integrating each gene expression data into a baseline transcriptome data set.

[0027] Optionally, the construction process of the blood routine data set comprises:

[0028] Obtaining health status blood routine data of a second test population, and second symptom condition data of the second test population related to acute upper respiratory tract infection in a historical period;

[0029] According to the second symptom condition data, the second test population is divided into a second susceptible group and a second non-susceptible group;

[0030] Performing inter-group comparison based on statistical difference analysis on the second susceptible group and the second non-susceptible group combined with the health status blood routine data, screening a plurality of blood routine indexes with statistical differences to form a blood routine index group affecting the susceptibility risk of acute upper respiratory tract infection in humans;

[0031] Extracting a plurality of blood routine index data corresponding to the blood routine index group from the health status blood routine data, and integrating each blood routine index data into a blood routine data set.

[0032] The application also provides a device for predicting the susceptibility of acute upper respiratory tract infection, which is applied to the method for predicting the susceptibility of acute upper respiratory tract infection; the method uses the prediction marker as described above to predict the susceptibility risk of acute upper respiratory tract infection in humans; and the device comprises:

[0033] A to-be-tested sample acquisition unit is configured to acquire a to-be-tested blood sample;

[0034] A blood analysis unit is configured to perform blood analysis based on the to-be-tested blood sample to obtain a blood routine sample and a gene expression sample;

[0035] The susceptible risk judgment unit is configured to input the blood routine sample and the gene expression sample into a susceptibility prediction model pre-trained based on the prediction marker to perform susceptible risk judgment, and obtain a susceptible risk evaluation result of the human body corresponding to the blood sample to be tested for acute upper respiratory tract infection in a future period of time.

[0036] The present application also provides an electronic device, which comprises a processor and a memory:

[0037] The memory is configured to store program code and transmit the program code to the processor.

[0038] The processor is configured to execute the method for predicting susceptibility to acute upper respiratory tract infection according to the instructions in the program code.

[0039] The present application also provides a computer readable storage medium for storing program code, which is used to execute the method for predicting susceptibility to acute upper respiratory tract infection.

[0040] From the above technical solutions, the present application has the following advantages:

[0041] In view of the deficiencies in the prior art, a prediction marker for predicting susceptibility to acute upper respiratory tract infection and a corresponding prediction method are provided. First, a prediction marker for predicting susceptibility to acute upper respiratory tract infection is provided, which is a prediction feature combination obtained by comprehensively considering the blood routine data and baseline blood transcriptome data of the human body and performing feature processing. Secondly, a prediction method is provided, which obtains a blood sample to be tested; performs blood analysis based on the blood sample to be tested, inputs the obtained blood routine sample and gene expression sample into a susceptibility prediction model pre-trained based on the prediction marker to perform susceptible risk judgment, and obtains a susceptible risk evaluation result of the human body corresponding to the blood sample to be tested for acute upper respiratory tract infection in a future period of time. Thus, by using the prediction marker for predicting susceptibility to acute upper respiratory tract infection and the corresponding prediction method, the susceptible risk of the human body for acute upper respiratory tract infection in a future period of time can be predicted simply, quickly and accurately. BRIEF DESCRIPTION OF DRAWINGS

[0042] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed in the embodiments or prior art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and those skilled in the art can obtain other drawings according to these drawings without creative labor.

[0043] Figure 1A schematic diagram of a construction and training process of a susceptibility prediction model;

[0044] Figure 2 A flowchart of steps of a method for predicting susceptibility of acute upper respiratory tract infection;

[0045] Figure 3 A structural block diagram of a device for predicting susceptibility of acute upper respiratory tract infection. DETAILED DESCRIPTION

[0046] The embodiment of the present application provides a prediction marker for susceptibility of acute upper respiratory tract infection, a prediction method for susceptibility of acute upper respiratory tract infection, a prediction device for susceptibility of acute upper respiratory tract infection, an electronic device and a storage medium, which are used to solve or partially solve the technical problem of low prediction accuracy of current susceptibility risk of acute upper respiratory tract infection.

[0047] In order to make the purposes, features and advantages of the present application more obvious and easy to understand, the technical solutions in the embodiments of the present application will be clearly and completely described below in combination with the drawings in the embodiments of the present application. Obviously, the following described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0048] As an example, the existing research on susceptibility of upper respiratory tract infection is mainly carried out through a virus challenge experiment. By inoculating a healthy volunteer with a rhinovirus, a respiratory syncytial virus, a coronavirus or an influenza virus, the infection and illness of the individual 4-10 days after inoculation are observed to explain the short-term susceptibility mechanism of a specific pathogen and to predict susceptibility. However, such research is limited by highly controlled experimental conditions and a short observation window of 4-10 days, and it is difficult to reflect the cumulative effects of dynamic exposure to multiple pathogens (in a certain period, each cold is infected with different pathogens, and it is not a single pathogen), continuous effects of environmental factors (changes in environmental temperature, humidity and air pollutant concentration), and long-term social and psychological factors of individuals (individual living habits, stress, sleep, diet, etc.) in the real environment, resulting in low prediction accuracy.

[0049] The embodiment of the present application further analyzes the currently used susceptibility prediction method, and summarizes the following points:

[0050] (1) At present, about 200 viruses and more than 10 common pathogenic bacteria can cause acute upper respiratory tract infection, but the current technology only focuses on specific single viruses, such as rhinovirus, influenza virus, respiratory syncytial virus and coronavirus, and lacks overall understanding of susceptibility to upper respiratory tract infection.

[0051] (2) Based on virus challenge experiments, only the susceptibility of individuals to a single virus can be captured in a short time (4-10 days), and the cumulative effects of changes in environmental exposure, dynamic exposure to multiple pathogens, and changes in individual social and psychological factors over a long period of time (such as one year) cannot be reflected.

[0052] (3) Most current technologies use data after infection to predict the severity of symptoms.

[0053] (4) Only a few studies focus on the association between immune characteristics at baseline or in a healthy state and susceptibility to upper respiratory tract infection, and transcriptome data are not fully utilized for predictive modeling.

[0054] (5) Currently, there is no study that uses blood routine data in a healthy state to predict the susceptibility of individuals to acute upper respiratory tract infection. Blood routine data is simple, fast, and cost-effective, and routine examination indicators have high stability and repeatability.

[0055] (6) Genome-wide GWAS (Genome-Wide Association Study) analysis can only reveal some genetic factors. The genome is static and cannot reflect the dynamic immune characteristics of individuals at different time points. Genome analysis requires large sample sizes in the order of ten thousand.

[0056] Among them, the transcriptome refers to the collection of all RNAs (Ribonucleic Acid) expressed by a specific cell or tissue under specific conditions. Transcriptome sequencing technology can capture the changes in gene expression profiles of individuals in disease or healthy states with high throughput, thereby revealing key molecular pathways, immune response types, and individual susceptibility characteristics to infection. In susceptibility studies, transcriptome data can be used to identify immune activation or inhibition signals related to viral infection, providing molecular evidence for individualized risk assessment and therapeutic intervention. Blood routine is one of the most commonly used basic clinical detection methods. By detecting various blood cells (such as white blood cells, red blood cells, and platelets) and their classification counts in peripheral blood, the immune status and infection response of individuals can be preliminarily evaluated. This method is fast, economical, and highly reproducible, and can be used to establish an infection risk prediction model.

[0057] Therefore, one of the core points of the embodiments of the present application is to combine the health state blood routine data and the baseline blood transcriptome data to construct a susceptibility prediction model in a population to predict the susceptibility of acute upper respiratory tract infection. First, a prediction marker for the susceptibility of acute upper respiratory tract infection is provided, which is a prediction feature combination obtained by comprehensively considering the health state blood routine data and the baseline blood transcriptome data of the human body and performing feature processing. Secondly, a corresponding prediction method for the susceptibility of acute upper respiratory tract infection is provided. The blood sample to be tested is obtained; the blood sample to be tested is analyzed based on the blood sample to be tested to obtain a blood routine sample and a gene expression sample; the blood routine sample and the gene expression sample are input into the susceptibility prediction model pre-trained based on the prediction marker to perform susceptibility risk judgment, and the susceptibility risk evaluation result of the human body corresponding to the blood sample to be tested in the future period of time is obtained. Thus, the susceptibility risk of the human body to acute upper respiratory tract infection is simply, quickly and accurately predicted. By using the technical solution of the present application, on the one hand, the prediction marker for predicting the susceptibility of acute upper respiratory tract infection in the blood of the human body is found based on the machine learning algorithm, and the susceptibility risk of the individual to acute upper respiratory tract infection is predicted in the baseline and health state. On the other hand, the individual heterogeneity in the clinical comprehensive syndrome of acute upper respiratory tract infection is concerned, the susceptibility prediction model is constructed based on the health state blood routine data and the baseline transcriptome data, has high prediction accuracy and good practicability, and can also make up for the deficiency of the current research which only focuses on the prediction of susceptibility to specific viruses. In addition, the blood routine index is easy to obtain and has low cost, and has the feasibility of practical promotion. The transcriptome data is accurately identified by the baseline immune characteristics, and provides molecular basis for individualized prevention. The combination of blood routine and transcriptome data is expected to realize the comprehensive evaluation of the susceptibility of acute upper respiratory tract infection with controllable cost and high precision, and promote the accurate implementation of individualized prevention and control strategies.

[0058] Based on the foregoing examples and core points, the present application mainly includes two aspects: first, a prediction marker for the susceptibility of acute upper respiratory tract infection is provided; second, a prediction method for accurately predicting the susceptibility of acute upper respiratory tract infection by the prediction marker is provided.

[0059] Regarding the first aspect, in specific implementation, the embodiments of the present application provide a prediction marker for the susceptibility of acute upper respiratory tract infection. The prediction marker is a prediction feature combination obtained by comprehensively considering the health state blood routine data and the baseline blood transcriptome data of the human body and performing feature processing; the prediction marker is used to predict the susceptibility risk of the human body to acute upper respiratory tract infection.

[0060] Further, the prediction feature combination includes a blood routine index group and a differential expression gene group. The blood routine index group is obtained based on statistical difference analysis of health state blood routine data. The differential expression gene group is obtained based on gene differential expression analysis and feature screening based on baseline blood transcriptome data.

[0061] Further, the blood routine index group includes RDW-CV (Red Cell Distribution Width -Coefficient of Variation), P-LCR (Platelet Large Cell Ratio), MONO (Monocyte Absolute Value), Hb (Hemoglobin), and MPV (Mean Platelet Volume). The differential expression gene group includes A2M, TBC1D30, MCIR, PHOSPHO1, TLCD4, B3GALNT1, TACSTD2, and ZNF154.

[0062] The prediction marker for susceptibility of acute upper respiratory tract infection provided in the embodiment of the present application will be determined and verified by machine learning algorithm based on the experimental data related to acute upper respiratory tract infection in the following examples.

[0063] Regarding the content of the second aspect, in a specific implementation, first, a susceptibility prediction model for judging the susceptibility risk of acute upper respiratory tract infection needs to be constructed. To better illustrate, Figure 1 A construction and training process schematic diagram of a susceptibility prediction model provided in the embodiment of the present application is shown.

[0064] Further, in combination with Figure 1 and specific experiments, the construction and training process of the susceptibility prediction model can be implemented by performing the following sub-steps S01 to S06:

[0065] Step S01: Obtain a pre-constructed blood routine dataset and a baseline transcriptome dataset;

[0066] The blood routine dataset corresponds to a blood routine index group with statistical difference. The baseline transcriptome dataset corresponds to a differential expression gene group with gene expression difference.

[0067] First, participant recruitment. In the period from March 2023 to April 2023, 166 healthy participants were included in the undergraduate and graduate student groups of the Shenzhen Campus of Sun Yat-sen University by the following criteria. The age range of the participants is 18-29 years old.

[0068] Before enrollment, the general condition of the participants was investigated. The content includes basic information, general health status, history of allergy, immunization history, smoking and drinking, frequency of previous respiratory tract infection, etc.

[0069] The specific inclusion criteria are as follows:

[0070] (1) Undergraduate and graduate students enrolled in Shenzhen Campus of Sun Yat-sen University after June 30, 2023.

[0071] (2) Low-grade students. Can be followed up for a long time. Healthy, no serious underlying diseases or chronic diseases. Age greater than 18 years old. Can understand and voluntarily participate in the study.

[0072] (3) Has signed the informed consent form. And agree to provide necessary biological samples (such as blood, throat test paper, etc.) and participate in questionnaire survey.

[0073] (4) Good compliance. No factors that hinder subsequent follow-up at the time of enrollment.

[0074] The specific exclusion criteria are as follows:

[0075] (1) Participants are pregnant or lactating women. Pregnancy or lactation may have a special impact on immune status or susceptibility.

[0076] (2) Participants have underlying diseases. Including chronic respiratory diseases, malignant tumors, cardiovascular and cerebrovascular diseases, metabolic diseases, blood diseases, autoimmune diseases, immunodeficiency diseases, severe gastrointestinal diseases, acute infectious diseases, hepatitis, etc.

[0077] (3) Participants have mental illness or cognitive impairment.

[0078] (4) Participants are unwilling to provide biological samples or unable to be followed up for a long time.

[0079] (5) Participants have participated in other intervention studies, which may affect the results of this study.

[0080] (6) Participants have a history of smoking or frequent drinking (≥3 days of drinking per week).

[0081] (7) Participants have used antibiotics more than 2 times per year for respiratory tract infection in the past 3 years. Including amoxicillin, cephalosporins, erythromycin, azithromycin, etc.

[0082] (8) Participants have obvious symptoms of acute infection at the time of enrollment. These participants need to be excluded to ensure that the immune baseline status is normal at the time of enrollment.

[0083] (9) Participants are members of the research team or family members.

[0084] (10) Participants failed to complete baseline transcriptome sample collection or sample detection after sample collection.

[0085] The follow-up time of the study was from the 29th week of 2023 (2023-07-17) to the 28th week of 2024 (2024-07-14). Since all participants were students of the Shenzhen Campus of Sun Yat-sen University, it was convenient to contact each participant and timely understand their message reading status. Therefore, the participants were managed by establishing a group. The researchers prepared a follow-up questionnaire every Monday and published it in the management group.

[0086] The participants filled in the online questionnaire about the upper respiratory tract infection in the past week. Specifically, whether there were upper respiratory tract symptoms in the past week. If so, further select whether there were symptoms such as dry throat, sore throat, cough, runny nose, nasal congestion, sneezing, fever, dizziness and headache, fatigue, muscle soreness, diarrhea, etc. according to the actual situation. The researchers analyzed and scored according to the participant's filling. If there were no upper respiratory tract symptoms, it was scored as 0. If there were upper respiratory tract symptoms, it was scored as 1, 2 or 3 according to the severity of the symptoms.

[0087] Among them, participants with or without upper respiratory tract infection had to fill out the questionnaire. The researchers updated the questionnaire name and guide every week to remind the participants of the time period of the questionnaire survey. During the follow-up period, there were 51 follow-ups.

[0088] The researchers took respiratory throat swab samples from 20% of the participants who reported upper respiratory tract infection to confirm the authenticity of the online questionnaire survey. The detection of rhinovirus, coronavirus, influenza virus, etc. through sampling proved the applicability of the invention to the susceptibility judgment of acute upper respiratory tract infection caused by multiple pathogens.

[0089] All participants were healthy young student population. When the study subjects were included, individuals who used antibiotics more than twice a year due to respiratory tract infection in the past three years were excluded. Considering the incidence of upper respiratory tract infection in healthy adults (2-4 times per year (Lancet, 2003), 2-5 times (Lancet Infect Dis, 2005), or 2-3 times (CMAJ, 2014)), and the sample size balance problem of subsequent model construction, the susceptible group was defined as participants who reported acute upper respiratory tract infection symptoms for ≥2 weeks during the follow-up period. If a participant reported acute upper respiratory tract infection symptoms for 0 or 1 week during the follow-up period, he / she was defined as not susceptible to upper respiratory tract infection.

[0090] For participants (59) whose weekly follow-up questionnaire filling rate was less than 80% (filling frequency was less than 41 times) during the study period, the number of weeks reporting upper respiratory tract infection during the follow-up period and the "average annual frequency of cold symptoms" filled in the questionnaire before enrollment were checked one by one, and the participants were contacted online for follow-up to determine the susceptibility grouping of the participants.

[0091] It should be noted that all the test data sources used in the related experiments in the embodiments of the present application are legal.

[0092] Further, the baseline transcriptome dataset proposed in the embodiments of the present application can be constructed by performing the following sub-steps S11 to S16:

[0093] Step S11: obtaining baseline blood transcriptome data of a first test population and first symptom condition data of the first test population related to acute upper respiratory tract infection in a historical period;

[0094] From April 10 to 27, 2023, 166 participants were drawn by 10 mL of peripheral blood as baseline samples by school health service center nursing staff (to distinguish from the actual population participating in blood routine sampling after subsequent screening, the participant population included in the baseline blood transcriptome data sampling is defined as the first test population, and the participant population actually included in the blood routine data sampling is defined as the second test population). The blood sampling time is every 7:00~9:00.

[0095] Step S12: dividing the first test population into a first susceptible group and a first non-susceptible group according to the first symptom condition data;

[0096] In the collected baseline blood transcriptome data, according to the aforementioned symptom condition analysis of the susceptible group and the non-susceptible group, the first non-susceptible group corresponding to the baseline blood transcriptome test has a total of 74 samples, and the first susceptible group has a total of 92 samples.

[0097] Step S13: preprocessing the baseline blood transcriptome data to obtain a gene expression matrix and a plurality of PBMC samples;

[0098] PBMC (Peripheral Blood Mononuclear Cells, peripheral blood mononuclear cells) samples of each participant test sample are extracted, and PE150bp sequencing (150bp paired-end sequencing, 150 base pair paired-end sequencing) is performed on each PBMC sample. After returning the sequencing data, the original sequencing data is preprocessed, specifically: using FastQC quality control and HISAT2 alignment; converting the sam format alignment result to bam format, and sorting and indexing it; using featureCounts for quantification to generate a gene expression matrix.

[0099] Step S14: screening protein-coding genes with an expression ratio greater than or equal to a preset expression ratio from the gene expression matrix, and normalizing the protein-coding genes by TPM standardization to obtain a target gene expression matrix;

[0100] The columns of the gene expression matrix are sample (participant) IDs, and the rows are genes. Genes with expression in at least 50% of the samples are retained, and GTF files are used for gene annotation to filter out protein-coding genes. The protein-coding genes (i.e., the transcriptome sequencing data after gene filtering) are normalized by the TPM (Transcripts Per Million) standardization method to eliminate the effects of sequencing depth and gene length on expression. This step is mainly completed by using the R language version 4.4.1, rtracklayer, GenomicRanges, edgeR, RSEM, and DESeq2 packages.

[0101] Step S15: Based on the target gene expression matrix, gene differential expression analysis is performed on the first susceptible group and the first non-susceptible group, and combined with feature screening, a plurality of differential expression genes meeting the preset expression change condition are screened out to form a differential expression gene group that has an influence on the risk of human acute upper respiratory tract infection susceptibility;

[0102] The R language DESeq2 package can be used for gene differential expression analysis to identify the gene expression differences between the first susceptible group and the first non-susceptible group. In this regard, DESeq2 uses Wald test to calculate P value, and Benjamini-Hochberg method is used for multiple hypothesis testing correction. The differential fold (Fold Change, FC) represents the ratio between the average expression of a certain gene in the experimental group (the first susceptible group) and the average expression of the gene in the control group (the first non-susceptible group). The differential expression change gene screening standard can be set as |log2FC|> 0.25 and P<0.05.

[0103] Through preliminary screening, 365 differential expression genes are identified in the baseline state. According to the order of P value from small to large in the gene differential expression analysis, the top 50 differential genes are selected. Then, the recursive feature elimination method is used for feature screening from the 50 differential genes. The number of samples in each class of the classification task is at least 5-10 times the number of features, so the model related to the baseline blood transcriptome in the embodiment of the application finally retains 8 differential expression genes as part of the prediction features, which are included in the prediction markers. Among them, the differential expression gene group includes A2M, TBC1D30, MCIR, PHOSPHO1, TLCD4, B3GALNT1, TACSTD2, and ZNF154.

[0104] Step S16: Extract the gene expression data of the differential expression gene group in each PBMC sample from the gene expression profile, and integrate each gene expression data into a baseline transcriptome dataset.

[0105] Further, the blood routine dataset according to the embodiment of the present application can be constructed by performing the following sub-steps S21 to S24.

[0106] Step S21: obtaining health state blood routine data of the second test population and second symptom condition data of the second test population related to acute upper respiratory tract infection in a historical period;

[0107] In April 2024, the peripheral blood cell count (blood routine) data of the participants were collected according to the research needs. The health state screening criteria were defined as the physiological state when there was no acute upper respiratory tract infection in the past 60 days, and there were no obvious clinical infection symptoms, normal body temperature, and no cold or other acute disease manifestations at the current time point. Professional nursing staff collected 2 mL of peripheral blood from the participants, and sent it to the hospital medical laboratory emergency group laboratory for detection within 2 hours according to the blood routine (five classification). After the detection results were returned, the researchers arranged the data according to the electronic version of the test report. After screening, a total of 122 participants' blood routine data were collected in the health state blood routine data collection stage. In order to ensure the accuracy and effectiveness of the research results, 5 individuals with obvious abnormal clinical significance of blood routine were excluded. Specifically, if extreme values such as red blood cell count, white blood cell count, or platelet count are outside the normal reference range, the sample is excluded from the blood routine analysis. After screening, a total of 117 participants' blood routine data were included in the subsequent analysis.

[0108] Step S22: dividing the second test population into a second susceptible group and a second non-susceptible group according to the second symptom condition data;

[0109] Among the collected health state blood routine data, according to the aforementioned symptom condition analysis of the susceptible group and the non-susceptible group, the second non-susceptible group corresponding to the health state blood routine test has a total of 53 samples, and the second susceptible group has a total of 64 samples.

[0110] Step S23: comparing the second susceptible group and the second non-susceptible group based on statistical difference analysis combined with the health state blood routine data, screening a plurality of blood routine indexes with statistical differences to form a blood routine index group that exists to affect the susceptibility risk of acute upper respiratory tract infection in the human body;

[0111] In the intergroup comparison, t-test and Mann-Whitney U test are used. The significance level can be set as P<0.05. After analysis, there are five blood routine indexes with statistical differences. Respectively: the red blood cell distribution width variation coefficient, the large platelet percentage, the absolute value of mononuclear cells and the mean platelet volume of the susceptible group are higher than those of the non-susceptible group, and the hemoglobin capacity is lower than that of the non-susceptible group. The five blood routine indexes are also included in the prediction markers. Therefore, the blood routine index group screened by the embodiment of the application includes RDW-CV (red blood cell distribution width variation coefficient), P-LCR (large platelet percentage), MONO (absolute value of mononuclear cells), Hb (hemoglobin) and MPV (mean platelet volume). See Table 1 below for details.

[0112] Table 1: Differences in blood routine of the first susceptible group and the first non-susceptible group

[0113]

[0114] Among them, RDW-CV, P-LCR, unit %; MONO unit ; Hb unit g / L; MPV unit fL.

[0115] Step S24: Extracting a plurality of blood routine index data corresponding to the blood routine index group from the health state blood routine data, and integrating each blood routine index data into a blood routine data set.

[0116] Step S02: Dividing the blood routine data set into a blood routine training set and a blood routine test set according to a preset division ratio, and dividing the baseline transcriptome data set into a baseline transcriptome training set and a baseline transcriptome test set;

[0117] The data set can be divided into a training set and a test set according to an 8:2 ratio.

[0118] Step S03: Merging the blood routine index group and the differential expression gene group to construct a fusion feature matrix;

[0119] In order to further improve the prediction accuracy and application feasibility, the embodiment of the application constructs a multi-modal fusion model of blood routine and baseline transcriptome comprehensive prediction as a susceptibility prediction model of acute upper respiratory tract infection. Specifically, the blood routine index group corresponding to the blood routine data and the differential expression gene group corresponding to the baseline blood transcriptome data are modeled by fusion, and a fusion feature matrix is constructed by feature merging. Among them, the blood routine data reflects the macroscopic immunity and physiological state of the individual, and the baseline blood transcriptome data reveals the immune regulation mechanism at the molecular level. The two complement each other and are helpful to construct a more comprehensive prediction model.

[0120] Step S04: combining the fusion feature matrix as the prediction feature group of the susceptibility risk prediction model, taking acute upper respiratory tract infection susceptibility as the target variable, to construct an initial susceptibility prediction model;

[0121] In actual application, the present application can construct the corresponding susceptibility prediction model according to the health state blood routine data, baseline blood transcriptome differential gene data, and the combined data of blood routine and transcriptome according to the actual situation.

[0122] Among them, for each application situation, a plurality of different prediction model frameworks can be selected to construct an initial prediction model, including Random Forest (RF), Extreme Gradient Boosting (XGBoost), Support Vector Machine (SVM), etc. By respectively training and testing each prediction model framework, the optimal prediction model with the best prediction result is selected as the final susceptibility prediction model for subsequent prediction application. For ease of understanding, only one of the optimal model frameworks is used for illustration in the embodiments of the present application.

[0123] For the construction and training of the blood routine susceptibility prediction model, the blood routine index group with statistical difference is taken as the prediction model feature, and the acute upper respiratory tract infection susceptibility is taken as the target variable to construct an initial prediction model. In the model training and evaluation, the blood routine data set (corresponding to 117 samples, of which 53 samples are not susceptible group and 64 samples are susceptible group) is randomly divided into training set and test set in the ratio of 8:2. StandardScaler function is used to standardize the training set and test set data, and the data is converted to a distribution with a mean of 0 and a standard deviation of 1. In the training set, the Synthetic Minority Oversampling Technique (SMOTE) is used to balance the sample size of the susceptible group and the not susceptible group. In the training phase, the grid search combined with five-fold cross-validation is used to optimize the model hyperparameters to achieve the optimal balance accuracy of the model in the training set. The feature and hyperparameter configuration of the optimal prediction model are shown in Table 2.

[0124] For the construction and training of the baseline blood transcriptome susceptibility prediction model, the differentially expressed gene set with gene expression differences is used as the prediction model feature, the acute upper respiratory tract infection susceptibility is used as the target variable, and the baseline transcriptome dataset (i.e., corresponding to 166 samples, of which 74 samples are in the non-susceptible group and 92 samples are in the susceptible group) is divided into a training set and a test set in a ratio of 8:2. Data standardization, class balancing, and model training refer to the blood routine susceptibility prediction model, which will not be repeated here. The features and hyperparameter configurations of the optimal prediction model are shown in Table 2.

[0125] For the multi-modal fusion model for blood routine and baseline transcriptome comprehensive prediction, the fusion feature matrix (containing both the blood routine index group and the differentially expressed gene group) is used as the prediction feature combination for the susceptibility risk prediction model, and the acute upper respiratory tract infection susceptibility is used as the target variable to build an initial susceptibility prediction model. The data preprocessing (standardization, class balancing, etc.) and modeling process refer to the previous embodiments, which will not be repeated here. Through training, support vector machine can be selected as the optimal prediction model. The features and hyperparameter configurations of the optimal prediction model are shown in Table 2.

[0126] Table 2: Prediction model features and model hyperparameter configurations

[0127]

[0128] Step S05: Perform susceptibility risk judgment training on the initial susceptibility prediction model based on the blood routine training set and the baseline transcriptome training set, and perform model parameter adjustment and optimization during the training process.

[0129] Step S06: Perform model prediction performance evaluation on the susceptibility prediction model obtained after training by using the blood routine test set and the baseline transcriptome test set.

[0130] In the embodiments of the present application, balanced accuracy, AUC, PR-AUC, recall rate, and specificity indicators are used to evaluate the performance of the prediction model. The performance of each prediction model is shown in Table 3:

[0131] Table 3: Performance of prediction models on independent test sets

[0132]

[0133] Among them, the support vector machine model in the susceptibility prediction model using health state blood routine data performs best. The prediction performance is as follows: balanced accuracy 0.88, AUC (Area Under Curve) value 0.88, PR-AUC (Precision-Recall Area Under Curve) value 0.78, recall rate 0.91, specificity 0.85, and is suitable for predicting the susceptibility of acute upper respiratory tract infection using blood routine data.

[0134] In the susceptibility prediction model using baseline blood transcriptome data, the random forest model performs best. The prediction performance is as follows: balanced accuracy 0.82, AUC value 0.85, PR-AUC value 0.88, recall rate 0.83, specificity 0.81, and is suitable for predicting the susceptibility of acute upper respiratory tract infection in the baseline state.

[0135] In the fusion susceptibility prediction model integrating health state blood routine data and baseline blood transcriptome data, the support vector machine model performs best. The prediction performance is as follows: balanced accuracy 0.88, AUC value 0.93, PR-AUC value 0.96, recall rate 0.92, and specificity 0.83. The prediction model can realize simultaneous analysis of molecules and clinical parameters through one-time collection of blood samples in actual scenarios, and provide a technical basis for screening the susceptibility of infection in a large number of healthy people.

[0136] The above three prediction models can effectively identify positive samples when processing unknown data, and also have good discrimination ability for negative samples.

[0137] It can be understood that the sub-models (blood routine susceptibility prediction model, baseline blood transcriptome susceptibility prediction model) and the multi-modal fusion model (blood routine and baseline transcriptome comprehensive prediction multi-modal fusion model) can be regarded as a complementary relationship. The reason is that blood routine index belongs to low-cost and easy-to-obtain clinical basic detection data, and can reflect the macro physiological state. While the transcriptome data reveals the microscopic molecular level immune activation and disease susceptibility mechanism. Both have complementary prediction mechanisms.

[0138] In actual application, the embodiments of the present application can support two use paths. When the data is incomplete, the blood routine prediction model or the baseline blood transcriptome model is used for independent prediction. When the data is complete, the multi-modal fusion model is preferentially used for comprehensive prediction to obtain a higher AUC value.

[0139] The embodiments of the present application construct a susceptibility prediction model for acute upper respiratory tract infection through experiments, which has the following advantages:

[0140] In one aspect, a prediction model based on blood routine difference indicators under healthy state is provided. By screening statistically significant blood routine difference indicators as model input features, a support vector machine prediction model is constructed, which can effectively distinguish the susceptibility of adult acute upper respiratory tract infection before infection. The prediction model has a balanced accuracy of 0.88, an AUC value of 0.88, a PR-AUC value of 0.78, a recall rate of 0.91, and a specificity of 0.85, and has high recall rate and specificity, which is suitable for early screening of high-risk groups. The prediction features of the blood routine susceptibility prediction model include red cell distribution width coefficient of variation (RDW-CV), large platelet percentage (P-LCR), absolute value of mononuclear cells (MONO), hemoglobin (Hb), and mean platelet volume (MPV).

[0141] In another aspect, a prediction model based on baseline blood transcriptome difference genes is provided. By using the transcriptome expression data of individuals in the baseline state for gene differential expression analysis, specific expression genes are screened for training a random forest model. The prediction model has a balanced accuracy of 0.82, an AUC value of 0.85, a PR-AUC value of 0.88, a recall rate of 0.83, and a specificity of 0.81. Thus, the prediction model can accurately predict the susceptibility risk of healthy people to upper respiratory tract infection within a certain time range (e.g., one year) in the future, and has good balanced accuracy and model generalization ability. The prediction features of the baseline blood transcriptome susceptibility prediction model include a combination of 8 differentially expressed genes, including A2M, TBC1D30, MCIR, PHOSPHO1, TLCD4, B3GALNT1, TACSTD2, and ZNF154.

[0142] Further, on the basis of the aforementioned prediction model, a prediction strategy that fuses multi-modal data is provided. The present application first proposes a strategy of combining blood routine and baseline blood transcriptome expression data for comprehensive analysis and modeling. By processing features of different data dimensions, the stability and applicability of the prediction model are improved, which can adapt to different data and scenarios. The prediction features of the multi-modal fusion prediction model include 13 features, including red cell distribution width coefficient of variation (RDW-CV), large platelet percentage (P-LCR), absolute value of mononuclear cells (MONO), hemoglobin (Hb), mean platelet volume (MPV), A2M, TBC1D30, MCIR, PHOSPHO1, TLCD4, B3GALNT1, TACSTD2, and ZNF154.

[0143] In addition, the embodiment of the present application also provides a longitudinal prospective cohort design based on a natural state population. The data used in the embodiment of the present application are derived from a prospective follow-up cohort under non-intervention conditions, thereby avoiding various limitations that may exist in a virus challenge experiment, ensuring the clinical applicability and real generalizability of the model, and having important transformation value.

[0144] Based on the susceptibility prediction model constructed in the foregoing embodiment, referring to Figure 2 , a step flowchart of a method for predicting susceptibility to acute upper respiratory tract infection provided by the embodiment of the present application is shown. The method uses the prediction marker provided in the foregoing embodiment to predict the susceptibility risk of a human body to acute upper respiratory tract infection, and the method can specifically include the following steps:

[0145] Step 201, obtaining a blood sample to be tested;

[0146] Step 202, performing blood analysis based on the blood sample to be tested to obtain a blood routine sample and a gene expression sample;

[0147] The blood analysis includes detection analysis for blood routine and gene differential expression analysis for a baseline blood transcriptome. Through the blood routine detection analysis, the blood routine sample can be obtained, and through the gene differential expression analysis for the baseline blood transcriptome, the gene expression sample can be obtained.

[0148] Step 203, inputting the blood routine sample and the gene expression sample into a susceptibility prediction model pre-trained based on the prediction marker to perform susceptibility risk judgment, and obtaining a susceptibility risk evaluation result of a human body corresponding to the blood sample to be tested to acute upper respiratory tract infection in a future period of time.

[0149] The blood routine sample and the gene expression sample are inputted into a susceptibility prediction model (a multi-modal fusion prediction model) pre-trained based on the prediction marker to perform susceptibility risk judgment, and a susceptibility risk evaluation result of a human body corresponding to the blood sample to be tested to acute upper respiratory tract infection in a future period of time (such as one year) is obtained. The susceptibility risk evaluation result is susceptible or not susceptible.

[0150] In addition, based on the blood routine susceptibility prediction model and the baseline blood transcriptome susceptibility prediction model constructed in the foregoing, the blood routine sample can also be inputted into the blood routine susceptibility prediction model to perform susceptibility risk judgment, and a corresponding susceptibility risk evaluation result is obtained. Or the gene expression sample can also be inputted into the baseline blood transcriptome susceptibility prediction model to perform susceptibility risk judgment, and a corresponding susceptibility risk evaluation result is obtained. Those skilled in the art can select a suitable susceptibility prediction means based on actual conditions. It can be understood that the present application does not limit this.

[0151] In the embodiment of the present application, based on the susceptibility prediction model constructed in the foregoing embodiment, a method for predicting the susceptibility of acute upper respiratory tract infection is provided. The prediction method has the following advantages:

[0152] (1) The prior art relies on virus challenge experiments, and the data source is limited to a specific virus infection process, thereby limiting the applicability and generalizability of the research results. The present application is based on healthy people in a natural state, and through longitudinal cohort follow-up, an upper respiratory tract infection susceptibility prediction model based on health status blood routine, baseline transcriptome and fusion of the two is constructed, fully reflecting the localization and real adaptability of the data source, and making up for the blank of existing research in healthy people.

[0153] (2) The present application focuses on the prediction of the susceptibility of upper respiratory tract infection at the level of clinical syndromes, rather than only for a certain virus. The constructed prediction model can capture the overall susceptibility characteristics of individuals under multiple pathogen exposure, thereby reflecting the risk level of individuals under the combined action of multiple infection factors in real life environment, and expanding the application range of the prediction model.

[0154] (3) In addition, the present application first proposes to use blood routine data in a healthy state for susceptibility prediction, which is more clinically applicable and feasible than relying only on high-cost and high-technical threshold baseline blood transcriptome data. In terms of baseline blood transcriptome prediction model, the present application also innovatively realizes the prediction of the risk of illness in the future period (such as within one year) in the baseline state, breaking through the limitation of previous research in short time window.

[0155] (4) The present application takes the practicality of the technical solution, the universality of the prediction dimension and the adaptability of the data source as the core advantages, and significantly improves the scientificity and generalizability of the identification of the susceptibility of acute upper respiratory tract infection.

[0156] The prediction model of the susceptibility of acute upper respiratory tract infection constructed in the embodiment of the present application, although using prospective cohort design, blood routine and transcriptome feature extraction, differential analysis and various machine learning methods to achieve the purpose of the application. However, in the technical implementation path, the implementation means used in the foregoing embodiment is not unique, and there are still many alternative schemes. Specifically, but not limited to, the following aspects:

[0157] I. Alternative research design scheme

[0158] In addition to the prospective longitudinal cohort design, historical retrospective electronic health record and biological sample bank data can also be used in some embodiments, combined with the frequency of previous upper respiratory tract infection and baseline samples, to establish a label data set for building a susceptibility prediction model. A multi-center cross-sectional design can also be used to indirectly build a susceptibility label by combining questionnaire surveys and biological sample collection.

[0159] II. Alternative sample types

[0160] In some embodiments, in addition to using PBMC transcriptome and blood routine indicators, saliva, nasopharyngeal swab or peripheral blood whole blood samples can be used instead of transcriptome data. Alternatively, biomarkers such as protein, metabolite or cytokine levels in plasma and serum can be used to extract susceptibility features.

[0161] III. Alternative modeling algorithms

[0162] In some embodiments, in addition to support vector machines, random forests, and extreme gradient boosting models, logistic regression, light gradient boosting machine (LightGBM), neural networks (DNN (Deep Neural Network), MLP (Multi-Layer Perceptron)), and ensemble learning methods such as stacking, bagging, and boosting can also be used for model training.

[0163] IV. Alternative label determination criteria

[0164] In some embodiments, the determination of susceptibility to acute upper respiratory tract infection can be based on different follow-up frequencies or evaluation indicators. For example, the "upper respiratory tract infection susceptibility" label can be defined using a continuous symptom score scale, anti-infective drug use records, number of days absent, frequency of hospital visits, etc., to form the target variable for model training.

[0165] The symptom score method is as follows: participants fill in the symptom scores (such as Jackson score, WURSS-21 scale) of sore throat, cough, nasal congestion, fever, etc. every day, and the susceptibility is grouped according to the number of days with cumulative scores exceeding the threshold.

[0166] The anti-infective drug use record is as follows: the frequency of self-reported or platform-recorded antibiotic and antiviral drug use by participants during the follow-up period is counted, and those who exceed the annual median are determined to be "high risk / high sensitivity" individuals.

[0167] Absenteeism and medical behavior are specifically: combined with school attendance records and hospital medical information, if frequent sick leave due to cold (such as annual absence > 5 days) or multiple outpatient medical records, it can be used as a reference index for "susceptibility".

[0168] The mixed rule judgment is specifically: the above-mentioned multiple indexes are combined to establish a scoring system (for example, cold symptom score x 0.5 + drug use frequency x 0.3 + absent days x 0.2), and a weighted scoring mechanism is used to judge whether the individual is "susceptible to upper respiratory infection".

[0169] V. Alternative transcriptome data processing and modeling process

[0170] In some embodiments, in order to achieve the prediction of the susceptibility of individuals to upper respiratory tract infection by baseline transcriptome data, some specific steps (such as the use of DESeq2, TPM standardization, recursive feature elimination, and the retention of 8 features) adopted by the present application are not the only technical path, and the purpose of the application can also be achieved by alternative solutions. The alternative solutions include but are not limited to:

[0171] (1) Alternative method for constructing gene expression matrix: In addition to TPM standardization, the number of fragments per kilobase million (FPKM), the number of counts per million (CPM), or the built-in standardization method in the DESeq2 package can be used to correct the expression amount to eliminate the influence of sequencing depth and gene length.

[0172] (2) Alternative method for gene screening and annotation: In addition to the annotation and protein coding gene screening by GTF file combined with rtracklayer and GenomicRanges package, biomaRt, Ensembl REST API, AnnotationHub or GENCODE (Genome-wide Encyclopedia of DNA Elements) annotation database can also be used for functional gene screening.

[0173] (3) Alternative method for differential expression analysis: In addition to DESeq2, differential expression analysis can also use edgeR, limma-voom, NOISeq, etc. as an alternative. The difference significance statistics can use different statistical tests such as Wald test, likelihood ratio test (LRT) or moderated t-test.

[0174] (4) Feature selection method replacement: In addition to recursive feature elimination, weighted gene co-expression network analysis (WGCNA), Boruta algorithm, LASSO regression (Least Absolute Shrinkage and Selection Operator Regression), mutual information score, SHAP (SHapley Additive exPlanations) value analysis, ReliefF algorithm, and minimum redundancy maximum relevance can also be used for feature selection and dimensionality reduction.

[0175] (5) Replacement of feature number setting strategy: In addition to controlling the model to retain 8 features according to the ratio of the number of samples to the number of features, cross-validation can also be used to automatically select the optimal number of features, or embedded models such as L1 regularized SVM or Tree-based methods can be used to automatically complete feature compression during modeling.

[0176] Referring to Figure 3 , a structural block diagram of a device for predicting the susceptibility of acute upper respiratory tract infection is shown, which is applied to the method for predicting the susceptibility of acute upper respiratory tract infection provided in any of the preceding embodiments, and the method uses the prediction markers provided in the preceding embodiments to predict the susceptibility risk of acute upper respiratory tract infection in humans. The device can specifically include:

[0177] A to-be-tested sample acquisition unit 301 is configured to acquire a to-be-tested blood sample.

[0178] A blood analysis unit 302 is configured to perform blood analysis based on the to-be-tested blood sample to obtain a blood routine sample and a gene expression sample.

[0179] A susceptibility risk judgment unit 303 is configured to input the blood routine sample and the gene expression sample into a susceptibility prediction model that is pre-trained based on the prediction markers to perform susceptibility risk judgment and obtain a susceptibility risk evaluation result of acute upper respiratory tract infection in a human corresponding to the to-be-tested blood sample within a future period of time.

[0180] In an optional embodiment, the device further includes a susceptibility prediction model construction and training unit, which specifically includes:

[0181] a data set obtaining unit, configured to obtain a pre-constructed blood routine data set and a baseline transcriptome data set; the blood routine data set corresponds to a group of blood routine indexes with statistical differences; the baseline transcriptome data set corresponds to a group of differential expression genes with gene expression differences;

[0182] a data set division unit, configured to divide the blood routine data set into a blood routine training set and a blood routine test set according to a preset division ratio, and divide the baseline transcriptome data set into a baseline transcriptome training set and a baseline transcriptome test set;

[0183] a fusion feature matrix construction unit, configured to merge the group of blood routine indexes and the group of differential expression genes to construct a fusion feature matrix;

[0184] an initial susceptibility prediction model construction unit, configured to combine the fusion feature matrix as a prediction feature group of a susceptibility risk prediction model, take acute upper respiratory tract infection susceptibility as a target variable, and construct an initial susceptibility prediction model;

[0185] a model training unit, configured to perform susceptibility risk judgment training on the initial susceptibility prediction model based on the blood routine training set and the baseline transcriptome training set, and perform model parameter adjustment and optimization in the training process;

[0186] a model test unit, configured to perform model prediction performance evaluation on the susceptibility prediction model obtained after training by using the blood routine test set and the baseline transcriptome test set.

[0187] In an optional embodiment, the device further comprises a baseline transcriptome data set construction unit; the baseline transcriptome data set construction unit specifically comprises:

[0188] a baseline blood transcriptome data obtaining unit, configured to obtain baseline blood transcriptome data of a first test population and first symptom condition data of the first test population related to acute upper respiratory tract infection in a historical period;

[0189] a first test population division unit, configured to divide the first test population into a first susceptible group and a first non-susceptible group according to the first symptom condition data;

[0190] a data preprocessing unit, configured to preprocess the baseline blood transcriptome data to obtain a gene expression matrix and a plurality of PBMC samples;

[0191] a target gene expression matrix construction unit, configured to screen out protein coding genes with a sample expression ratio greater than or equal to a preset expression ratio from the gene expression matrix, and perform normalization processing on the protein coding genes by TPM standardization to obtain a target gene expression matrix;

[0192] The differential expression gene set determination unit is configured to perform gene differential expression analysis on the first susceptible group and the first non-susceptible group based on the target gene expression matrix, and perform feature screening to screen a plurality of differential expression genes meeting preset expression change conditions to form a differential expression gene set that has an influence on the human acute upper respiratory tract infection susceptibility risk.

[0193] The baseline transcriptome dataset integration unit is configured to extract gene expression data of the differential expression gene set in each PBMC sample from the gene expression profile, and integrate each gene expression data into a baseline transcriptome dataset.

[0194] In an optional embodiment, the device further comprises a blood routine dataset construction unit, and the blood routine dataset construction unit specifically comprises:

[0195] The healthy state blood routine data acquisition unit is configured to acquire healthy state blood routine data of a second test population and second symptom condition data of the second test population related to acute upper respiratory tract infection in a historical period.

[0196] The second test population division unit is configured to divide the second test population into a second susceptible group and a second non-susceptible group according to the second symptom condition data.

[0197] The blood routine index set determination unit is configured to perform inter-group comparison based on statistical difference analysis on the second susceptible group and the second non-susceptible group by combining the healthy state blood routine data, and screen a plurality of blood routine indexes having statistical difference to form a blood routine index set that has an influence on the human acute upper respiratory tract infection susceptibility risk.

[0198] The blood routine dataset integration unit is configured to extract a plurality of blood routine index data corresponding to the blood routine index set from the healthy state blood routine data, and integrate each blood routine index data into a blood routine dataset.

[0199] For the device embodiment, it is basically similar to the method embodiment, so it is described more simply, and the related parts refer to the foregoing method embodiment.

[0200] It should be noted that, in order to enable those skilled in the art to better distinguish data of the same type but different actual meanings, part of the technical features in the embodiments of the present application are distinguished by first and second, and the first and second are only used for data distinction, and have no other special meanings. It can be understood that the present application does not limit this.

[0201] The embodiments of the present application also provide an electronic device, and the device comprises a processor and a memory:

[0202] The memory is configured to store the program code and transmit the program code to the processor.

[0203] The processor is configured to execute the method for predicting the susceptibility of acute upper respiratory tract infection according to the instructions in the program code.

[0204] The embodiment of the present application further provides a computer readable storage medium, which is configured to store program code, and the program code is configured to execute the method for predicting the susceptibility of acute upper respiratory tract infection.

[0205] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the above-described system, device and unit can refer to the corresponding process in the foregoing method embodiment, and will not be repeated here.

[0206] In the several embodiments of the present application, it should be understood that the disclosed system, device and method can be implemented in other ways. For example, the device embodiments described above are only schematic. The division of the units is only a logical function division. There can be another division manner for actual implementation. For example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed mutual couplings or direct couplings or communication connections between the units can be indirect couplings or communication connections through some interfaces, devices or units, and can be electrical, mechanical or in other forms.

[0207] The units described as separate components can or can not be physically separate, and the components shown as units can or can not be physical units, that is, they can be located in one place, or can be distributed on a plurality of network units. According to actual needs, some or all of the units can be selected to achieve the purpose of the embodiment.

[0208] In addition, each functional unit in each embodiment of the present application can be integrated into a processing unit, or each unit can exist physically, or two or more units can be integrated into one unit. The integrated unit can be realized in the form of hardware, or in the form of a software functional unit.

[0209] The integrated unit, if implemented in the form of a software function unit and sold or used as an independent product, can be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the present application or the entire or part of the technical solutions that essentially contribute to the prior art can be embodied in the form of a software product. The computer software product is stored in a storage medium, and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in the various embodiments of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and various media that can store program codes.

[0210] The above description and the above embodiments are only used to illustrate the technical solutions of the present application, and not to limit them. Although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacements for some technical features. These modifications or replacements do not make the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A predictive marker for susceptibility to acute upper respiratory tract infection, characterized in that, The prediction marker is a prediction feature combination obtained by comprehensively considering the health state blood routine data and baseline blood transcriptome data and performing feature processing; the prediction marker is used for predicting the acute upper respiratory tract infection susceptibility risk of the human body.

2. The prognostic marker of claim 1, wherein, The prediction feature combination includes a blood routine index group and a differential expression gene group; the blood routine index group is obtained by performing statistical difference analysis based on the health state blood routine data; and the differential expression gene group is obtained by performing gene differential expression analysis and feature screening based on the baseline blood transcriptome data.

3. The prognostic marker of claim 2, wherein, The blood routine index group includes RDW-CV, P-LCR, MONO, Hb and MPV; and the differential expression gene group includes A2M, TBC1D30, MCIR, PHOSPHO1, TLCD4, B3GALNT1, TACSTD2 and ZNF154.

4. A method for predicting susceptibility to acute upper respiratory tract infection, characterized by, The method adopts the prediction marker in any one of claims 1 to 3 to predict the acute upper respiratory tract infection susceptibility risk of the human body; and the method includes: acquiring a to-be-tested blood sample; performing blood analysis based on the to-be-tested blood sample to obtain a blood routine sample and a gene expression sample; inputting the blood routine sample and the gene expression sample into a susceptibility prediction model pre-trained based on the prediction marker to perform susceptibility risk judgment, and obtaining a susceptibility risk evaluation result of the human body corresponding to the to-be-tested blood sample in a future period of time in terms of the acute upper respiratory tract infection.

5. The method of claim 4, wherein the acute upper respiratory infection susceptibility is predicted by the step of: The construction and training process of the susceptibility prediction model includes: ​ acquiring a pre-constructed blood routine dataset and a baseline transcriptome dataset; the blood routine dataset corresponds to a blood routine index group with statistical differences; and the baseline transcriptome dataset corresponds to a differential expression gene group with gene expression differences; dividing the blood routine dataset into a blood routine training set and a blood routine test set according to a preset division ratio, and dividing the baseline transcriptome dataset into a baseline transcriptome training set and a baseline transcriptome test set; merging the blood routine index group and the differential expression gene group to construct a fusion feature matrix; taking the fusion feature matrix as a prediction feature combination of a susceptibility risk prediction model, taking acute upper respiratory tract infection susceptibility as a target variable, and constructing an initial susceptibility prediction model; performing susceptibility risk judgment training on the initial susceptibility prediction model based on the blood routine training set and the baseline transcriptome training set, and performing model parameter adjustment and optimization during the training process; evaluating the model prediction performance of the susceptibility prediction model obtained after training by using the blood routine test set and the baseline transcriptome test set.

6. The method of predicting susceptibility to acute upper respiratory tract infection according to claim 5, characterized in that, The construction process of the baseline transcriptome dataset includes: acquiring baseline blood transcriptome data of a first test population and first symptom condition data of the first test population related to acute upper respiratory tract infection in a historical period; dividing the first test population into a first susceptible group and a first non-susceptible group according to the first symptom condition data; performing preprocessing on the baseline blood transcriptome data to obtain a gene expression matrix and a plurality of PBMC samples; Screening protein-coding genes with expression ratios greater than or equal to a preset expression ratio from the gene expression matrix, and normalizing the protein-coding genes by TPM standardization to obtain a target gene expression matrix; Based on the target gene expression matrix, gene differential expression analysis is performed on the first susceptible group and the first non-susceptible group, and combined with feature screening, a plurality of differentially expressed genes meeting the preset expression change condition are screened out to form a differentially expressed gene group that has an impact on the susceptibility risk of acute upper respiratory tract infection in humans; Gene expression data of the differentially expressed gene group in each PBMC sample is extracted from the gene expression profile, and each gene expression data is integrated into a baseline transcriptome dataset.

7. The method of claim 5, wherein the acute upper respiratory infection susceptibility is predicted by the step of: The construction process of the blood routine dataset includes: ​ Obtaining health status blood routine data of a second test population and second symptom condition data related to acute upper respiratory tract infection of the second test population within a historical period; According to the second symptom condition data, the second test population is divided into a second susceptible group and a second non-susceptible group; Combined with the health status blood routine data, inter-group comparison based on statistical difference analysis is performed on the second susceptible group and the second non-susceptible group, and a plurality of blood routine indicators with statistical differences are screened out to form a blood routine indicator group that has an impact on the susceptibility risk of acute upper respiratory tract infection in humans; A plurality of blood routine indicator data corresponding to the blood routine indicator group are extracted from the health status blood routine data, and each blood routine indicator data is integrated into a blood routine dataset.

8. A device for predicting susceptibility to acute upper respiratory tract infection, characterized by The device is applied to a prediction method of acute upper respiratory tract infection susceptibility; the method predicts the susceptibility risk of acute upper respiratory tract infection in humans by using the prediction marker of any one of claims 1 to 3; and the device includes: A to-be-tested sample acquisition unit configured to acquire a to-be-tested blood sample; A blood analysis unit configured to perform blood analysis based on the to-be-tested blood sample to obtain a blood routine sample and a gene expression sample; A susceptibility risk judgment unit configured to input the blood routine sample and the gene expression sample into a susceptibility prediction model pre-trained based on the prediction marker to perform susceptibility risk judgment, and obtain a susceptibility risk evaluation result of acute upper respiratory tract infection of a human body corresponding to the to-be-tested blood sample within a future period of time.

9. An electronic device, comprising: The device includes a processor and a memory: The memory is configured to store program code and transmit the program code to the processor; The processor is configured to execute the prediction method of acute upper respiratory tract infection susceptibility according to the instructions in the program code.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium is configured to store program code for executing the prediction method of acute upper respiratory tract infection susceptibility according to any one of claims 4 to 7.

Images