Computer device for predicting glucocorticoid treatment early drug response of ITP patient

Abstract

The invention discloses a computer device for predicting glucocorticoid treatment early drug response of ITP patients. The invention provides a computer device, which comprises a memory, a processor and a computer program stored on the memory, and the processor executes the computer program to realize the following steps: S1, receiving data: receiving sample data, the sample data are relative abundance of 10 kinds of level microorganisms, 3 kinds of alpha diversity indexes and 6 clinical indexes in an in-vitro fecal sample of a subject; s2, inputting data: inputting the sample data into an ITP glucocorticoid treatment drug response prediction model; s3, outputting a result: outputting the probability value of the subject through a prediction model; and obtaining the reaction effect of the ITP glucocorticoid treatment drug of the subject based on the probability value. The glucocorticoid treatment response of the ITP patient is predicted through the model, and a new thought is provided for clinical treatment decision of ITP.

Description

Technical Field

[0001] The invention belongs to the field of computer information and relates to a computer device for predicting early drug response of glucocorticoid treatment in ITP patients. Background Art

[0002] Immune thrombocytopenia (ITP) is an acquired organ-specific autoimmune disease, accounting for 30% of bleeding diseases and one of the most common bleeding diseases that seriously endangers human health. Glucocorticoids are the first-line treatment for ITP patients recommended by recognized guidelines. However, approximately 30% of ITP patients have a poor response to glucocorticoid treatment, and 50-85% of patients who are effective in hormone treatment still relapse in the first year of treatment. Further deterioration of relapsed / resistant patients, prolonged course of disease, switching of multiple treatment regimens and long-term use cause significant toxicity and are prone to develop potentially life-threatening bleeding events.

[0003] Therefore, it is necessary to find substances that can predict the drug response effect of glucocorticoids in the treatment of ITP so as to adjust the clinical treatment plan in a timely manner. Summary of the invention

[0004] The technical problem solved by the present invention is to provide a computer device for predicting the early drug response of ITP patients to glucocorticoid treatment.

[0005] In order to solve the above technical problem, the first aspect of the present invention provides a computer device, which includes a memory, a processor and a computer program stored in the memory, and the processor executes the computer program to implement the following steps:

[0006] S1. Receive data: Receive sample data, which includes 10 relative abundances of microorganisms at the species level, 3 alpha diversity indices, and 6 clinical indicators in the ex vivo fecal samples of the subjects;

[0007] The 10 species-level microorganisms are as follows: Bacteroides xylanisolvens, Bacteroides sp. 3_1_23, Bacteroides ovatus, Bacteroides sp. D2, Ruminococcus faecis, Parabacteroides gordonii, Bacteroides sp. 2_2_4, Turicibacter sanguinis, Desulfovibrio sp. An276 and Megamonas rupellensis;

[0008] The three alpha diversity indices are Chao1 index, Dominance index and Shannon index;

[0009] The six clinical indicators include gender, age, disease duration, platelet count before glucocorticoid treatment, World Health Organization bleeding score before glucocorticoid treatment, and whether glucocorticoid treatment was received in the past;

[0010] S2. Input data: input the sample data into the prediction model of glucocorticoid treatment drug response effect in ITP patients;

[0011] The glucocorticoid treatment drug response effect prediction model for ITP patients is constructed according to a method comprising the following steps: using the following data in in vitro fecal samples of ITP patients with known glucocorticoid treatment drug response and ITP patients with known glucocorticoid treatment drug non-response and the drug response or drug non-response corresponding to each sample as training samples to train the glucocorticoid treatment drug response effect model for ITP patients, wherein the following data are: the relative abundance of the 10 species-level microorganisms, the 3 alpha diversity indices and the 6 clinical indicators;

[0012] The glucocorticoid treatment drug response effect of the ITP patient is drug response or drug non-response;

[0013] S3, outputting the result: outputting the probability value of the predicted glucocorticoid treatment drug response effect of the subject through the glucocorticoid treatment drug response effect model of ITP patients; and then calculating the glucocorticoid treatment drug response effect of the subject based on the probability value;

[0014] The subject is an ITP patient.

[0015] In a second aspect, the present invention provides a computer program product, comprising a computer program, characterized in that when the computer program is executed by a processor, the steps in the computer device described in the first aspect are implemented.

[0016] In a third aspect, the present invention provides a computer-readable storage medium storing a computer program, wherein the computer program enables a computer to execute the steps in the computer device according to the first aspect.

[0017] In a fourth aspect, the present invention provides a device for predicting the drug response effect of glucocorticoids in the treatment of ITP, the device comprising:

[0018] S1, data receiving module: used to receive sample data, wherein the sample data is the relative abundance of 10 species-level microorganisms, 3 alpha diversity indices and 6 clinical indicators in the ex vivo fecal samples of the subjects;

[0019] The 10 species-level microorganisms are as follows: Bacteroides xylanisolvens, Bacteroides sp. 3_1_23, Bacteroides ovatus, Bacteroides sp. D2, Ruminococcus faecis, Parabacteroides gordonii, Bacteroides sp. 2_2_4, Turicibacter sanguinis, Desulfovibrio sp. An276 and Megamonas rupellensis;

[0020] The three alpha diversity indices are Chao1 index, Dominance index and Shannon index;

[0021] The six clinical indicators include gender, age, disease duration, platelet count before glucocorticoid treatment, World Health Organization bleeding score before glucocorticoid treatment, and whether glucocorticoid treatment was received in the past;

[0022] S2, data input module: used to input the sample data into the prediction model of glucocorticoid treatment drug response effect in ITP patients;

[0023] The glucocorticoid treatment drug response effect prediction model for ITP patients is constructed according to a method comprising the following steps: using the following data in in vitro fecal samples of ITP patients with known glucocorticoid treatment drug response and ITP patients with known glucocorticoid treatment drug non-response and the drug response or drug non-response corresponding to each sample as training samples to train the glucocorticoid treatment drug response effect model for ITP patients, wherein the following data are: the relative abundance of the 10 species-level microorganisms, the 3 alpha diversity indices and the 6 clinical indicators;

[0024] The glucocorticoid treatment drug response effect of the ITP patient is drug response or drug non-response;

[0025] S3, result output module: used for outputting a probability value of predicting the glucocorticoid treatment drug response effect of the subject through the glucocorticoid treatment drug response effect model of the ITP patient; and then calculating the glucocorticoid treatment drug response effect of the subject based on the probability value;

[0026] The subject is an ITP patient.

[0027] In a fifth aspect, the present invention provides a method for predicting or assisting in predicting a subject's glucocorticoid drug response effect in treating ITP, the method comprising the following steps:

[0028] S1. Obtain the relative abundance of 10 species-level microorganisms, 3 alpha diversity indices and 6 clinical indicators in the subjects' in vitro fecal samples;

[0029] The 10 species-level microorganisms are as follows: Bacteroides xylanisolvens, Bacteroides sp. 3_1_23, Bacteroides ovatus, Bacteroides sp. D2, Ruminococcus faecis, Parabacteroides gordonii, Bacteroides sp. 2_2_4, Turicibacter sanguinis, Desulfovibrio sp. An276 and Megamonas rupellensis;

[0030] The three alpha diversity indices are Chao1 index, Dominance index and Shannon index;

[0031] The six clinical indicators include gender, age, disease duration, platelet count before glucocorticoid treatment, World Health Organization bleeding score before glucocorticoid treatment, and whether glucocorticoid treatment was received in the past;

[0032] S2, inputting the sample data into a prediction model for the effect of glucocorticoid treatment on ITP patients;

[0033] The glucocorticoid treatment drug response effect prediction model for ITP patients is constructed according to a method comprising the following steps: using the following data in in vitro fecal samples of ITP patients with known glucocorticoid treatment drug response and ITP patients with known glucocorticoid treatment drug non-response and the drug response or drug non-response corresponding to each sample as training samples to train the glucocorticoid treatment drug response effect model for ITP patients, wherein the following data are: the relative abundance of the 10 species-level microorganisms, the 3 alpha diversity indices and the 6 clinical indicators;

[0034] The glucocorticoid treatment drug response effect of the ITP patient is drug response or drug non-response;

[0035] S3, predicting the probability value of the subject's predicted glucocorticoid treatment drug response effect through the output of the glucocorticoid treatment drug response effect model for ITP patients; and then calculating the glucocorticoid treatment drug response effect of the subject based on the probability value;

[0036] The subject is an ITP patient.

[0037] The method described above only includes data processing steps and does not include steps for processing physical samples, such as steps for obtaining cells, tissues and / or organs from a living human or animal body, steps for processing in vitro samples from a living human or animal body, or steps for processing a living human or animal body.

[0038] In a sixth aspect, the present invention provides a method for constructing a prediction model for the effect of glucocorticoid therapy on ITP patients, the method comprising the following steps: using the following data in in vitro fecal samples of ITP patients with known glucocorticoid therapy drug responses and ITP patients with known glucocorticoid therapy drug non-response and the drug response or drug non-response corresponding to each sample as training samples to train the glucocorticoid therapy drug response effect model for ITP patients, the following data being: the relative abundance of the 10 species-level microorganisms, the 3 alpha diversity indices and the 6 clinical indicators;

[0039] The glucocorticoid treatment drug response effect of the ITP patient is drug response or drug non-response.

[0040] The relative abundance of the above microorganisms = (the number of 16S rRNA sequences of each species level microorganism / the total number of 16S rRNA sequences in the sample)%.

[0041] The 16S rRNA is detected by any one or more of metagenomic sequencing, 16S-rRNA sequencing, ITS sequencing, qRT-PCR, Southern blotting, and in situ hybridization to obtain the number of 16S rRNA sequences.

[0042] The above three alpha diversity indices can be derived from metagenomic sequencing or 16S-rRNA sequencing results of in vitro fecal samples of subjects.

[0043] The Chao1 index is a nonparametric statistical method used to estimate the number of unobserved species (species richness) in a community. The larger the value, the higher the species richness in the community.

[0044] The Dominance index is an indicator to measure the degree to which a species or a few species dominate a community (species evenness). The closer its value is to 1, the more dominant a species or a few species are.

[0045] The Shannon index is an index for measuring community diversity, which takes into account species richness and uniformity. It is also called the species diversity index or information entropy index. The larger the value, the higher the diversity of the community.

[0046] The Chao1 index, Dominance index and Shannon index were calculated using the R language vegan package (see Oksanen, Jari, et al. "vegan: Community Ecology Package." R package, version 2.0-10, 2013. https: / / CRAN.R-project.org / package=vegan.);

[0047] The calculation formula of the Chao1 index is shown in Formula 1:

[0048] S Chao1 =S obs +2F1+2F2 (Formula 1);

[0049] Among them, S obs is the number of all species at the species level in the metagenomic sequencing results in the sample, F1 is the number of species that appear once at the species level in the metagenomic sequencing results in the sample, and F2 is the number of species that appear twice at the species level in the metagenomic sequencing results in the sample;

[0050] The calculation formula of the Dominance index is shown in Formula 2:

[0051]

[0052] Among them, S obs is the number of all species at the species level in the metagenomic sequencing results of the sample, p i It is the ratio of the number of individuals of the i-th species level to the total number of individuals of all species at the species level;

[0053] The calculation formula of the Shannon index is shown in Formula 3:

[0054]

[0055] Among them, S obs is the number of all species at the species level in the metagenomic sequencing results of the sample, p iIt is the ratio of the number of individuals of the i-th species level to the total number of individuals of all species at the species level.

[0056] In the above, the glucocorticoid treatment ITP drug response effect of the subject is the early drug response effect of the glucocorticoid treatment of the subject.

[0057] The glucocorticoid drug response effect of the subject in treating ITP is the glucocorticoid drug response effect of the subject in treating ITP for 1 month.

[0058] The subjects are ITP patients, and the ITP patients are all ITP patients, including newly diagnosed / persistent ITP patients and chronic ITP patients.

[0059] The present invention collects baseline stool samples from ITP patients, uses metagenomic sequencing to detect the composition of intestinal flora, and predicts the first-line glucocorticoid treatment response of ITP patients through an integrated learning model, providing new ideas for clinical treatment decision-making of ITP. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 Flowchart of the model for predicting the drug response to glucocorticoid therapy in patients with ITP.

[0061] Figure 2 The abundance and diversity information of bacterial species is collected for training.

[0062] Figure 3 is the training set ROC.

[0063] Figure 4 is the validation set ROC. DETAILED DESCRIPTION

[0064] The present invention is further described in detail below in conjunction with specific embodiments, and the examples provided are only for illustrating the present invention, rather than for limiting the scope of the present invention. The examples provided below can be used as a guide for further improvements by those of ordinary skill in the art, and do not constitute a limitation of the present invention in any way.

[0065] The experimental methods in the following examples, unless otherwise specified, are all conventional methods, and are performed according to the techniques or conditions described in the literature in the field or according to the product instructions. The materials, reagents, etc. used in the following examples, unless otherwise specified, can all be obtained from commercial channels.

[0066] Unless otherwise specified, the quantitative tests in the following examples were performed three times and the results were averaged.

[0067] The microorganisms in the following examples are as follows:

[0068] Bacteroides xylanisolvens is described in the literature “Chassard C, Delmas E, Lawson PA, Bernalier-Donadille A. Bacteroides xylanisolvens sp. nov., a xylan-degrading bacterium isolated from human faeces. Int J Syst Evol Microbiol. 2008; 58 (Pt

[0069] 4):1008-1013.doi:10.1099 / ijs.0.65504-0”.

[0070] Bacteroides sp.3_1_23 is described in the document “Liu NN, Duan XL, Ai X, Yang YT, LiM, Dou SX, Rety S, Deprez E, Xi XG. The Bacteroides sp.3_1_23 Pif1 protein is a multifunctional helicase. Nucleic Acids Res. 2015 Oct 15; 43(18): 8942-54. doi: 10.1093 / nar / gkv916”.

[0071] Bacteroides ovatus is described in the document “Booth SJ, Johnson JL, Wilkins TD. Bacteriocin production by strains of Bacteroides isolated from human feces and the role of these strains in the bacterial ecology of the colon. Antimicrob Agents Chemother. 1977; 11(4): 718-724. doi: 10.1128 / AAC.11.4.718”.

[0072] Bacteroides sp.D2 is recorded in the literature "Schoch CL, Ciufo S, Domrachev M, Hotton CL, Kannan S, Khovanskaya R, Leipe D, Mcveigh R, O'Neill K, Robbertse B, Sharma S, Soussov V, Sullivan JP, Sun L, Turner S, Karsch-Mizrachi I. NCBITaxonomy: a comprehensive update on curation, resources and tools.Database(Oxford).2020Jan 1;2020:baaa062.doi:10.1093 / database / baaa062".

[0073] Ruminococcus faecis is described in the document “Kim MS, Roh SW, Bae JW. Ruminococcus faecis sp. nov., isolated from human faeces. J Microbiol. 2011; 49(3): 487-491. doi: 10.1007 / s12275-011-0505-7”.

[0074] Parabacteroides gordonii is described in the document “Sakamoto M, Suzuki N, Matsunaga N, et al. Parabacteroides gordonii sp. nov., isolated from human blood cultures. Int J Syst Evol Microbiol. 2009; 59(Pt 11): 2843-2847. doi: 10.1099 / ijs.0.010611-0”.

[0075] [ PubMed ] Schoch CL,Ciufo S,Domrachev M,Hotton CL,Kannan S,Khovanskaya R,Leipe D,Mcveigh R,O'Neill K,Robbertse B,Sharma S,Soussov V,Sullivan JP,Sun L,Turner S,Karsch-Mizrachi I.NCBITaxonomy:a comprehensive update on curation,resources and tools.Database(Oxford).

[0076] Interactive Turicibacter sanguinis gen.nov.,sp.nov.,a novel anaerobic,Gram-positive bacterium.Int J Syst Evol Microbiol.2002:52(Pt 4):1263–1266.

[0077] doi:10.1099 / 00207713-52-4-1263” .

[0078] [ PMC free article ] [ PubMed ] Schoch CL,Ciufo S,DomrachevM,Hotton CL,Kannan S,Khovanskaya R,Leipe D,Mcveigh R,O'Neill K,Robbertse B,Sharma S,Soussov V,Sullivan JP,Sun L,Turner S,Karsch-Mizrachi I.NCBITaxonomy:a comprehensive update on curation,resources and tools.Database(Oxford).

[0079] Megamonas rupellensis is described in “Chevrot R, Carlotti A, Sopena V, Marchand P, Rosenfeld E. Megamonas rupellensis sp. nov., an anaerobe isolated from the caecum of a duck. Int J Syst Evol Microbiol. 2008; 58(Pt 12): 2921-2924. doi: 10.1099 / ijs.0.2008 / 001297-0”.

[0080] Subject enrollment and clinical information collection in the following examples:

[0081] All subjects who met the diagnostic criteria for immune thrombocytopenia and whose diagnosis was confirmed and whose physicians or patients chose to take the first-line treatment (glucocorticoid therapy, which may be accompanied by platelet transfusion) signed informed consent.

[0082] Training set: 100 ITP patients who met the ITP diagnostic criteria in the outpatient department of Peking University People's Hospital from March to September 2018 and April to November 2021 were consecutively included.

[0083] Validation set: 52 ITP patients admitted to hematology outpatient clinics or wards from 20 medical centers across the country from October 2021 to January 2022 were consecutively included.

[0084] Subject inclusion criteria: isolated thrombocytopenia (platelet count <100×10 9 / L), aged >14 years, were diagnosed with ITP according to the internationally recognized ITP diagnostic criteria (The American Society of Hematology 2011evidence-basedpracticeguideline for immune thrombocytopenia.Blood.2011;117(16):4190-207.). Routine blood tests showed normal white blood cells and red blood cells, and no medication for thrombocytopenia was received within 6 months. At the time of initial diagnosis, a comprehensive medical history, physical examination, blood cell count, and peripheral blood film analysis were performed. Helicobacter pylori, hepatitis B virus (HBV), human immunodeficiency virus (HIV), and hepatitis C virus (HCV) infections, blood antinuclear antibodies, antiphospholipid antibodies, antithyroid antibodies, and thyroid hormone levels were tested for adult patients suspected of ITP.

[0085] Subject exclusion criteria: exclude secondary ITP such as drug-related thrombocytopenia; viral infection causing thrombocytopenia (HIV, HBV or HCV); severe heart, kidney, liver or respiratory insufficiency; severe immunodeficiency; pregnancy or lactation; myelodysplastic disease or myelofibrosis; history of malignant tumors; receiving immunosuppressive therapy for other diseases. If the patient has taken any medication for thrombocytopenia within 6 months, or has been exposed to any antibiotics, probiotics or probiotics within 4 weeks, or is receiving immunosuppressive therapy before stool sampling, he / she will not be included in this study.

[0086] Clinical indicators included gender, age, disease duration (in months), platelet count, World Health Organization bleeding score (Reporting results of cancer treatment. Cancer. 1981; 47(1): 207-214.), and previous treatment with glucocorticoids.

[0087] According to the international consensus on the treatment of ITP, the subjects of this study received standard glucocorticoid treatment and the platelet count was tested every 7 days. The treatment options included the following 2 categories: ① Prednisone or methylprednisolone: ​​1 mg (prednisone) or 0.8 mg (methylprednisolone) per kilogram of body weight per day, divided or taken all at once, reduced after the effect, and discontinued within 6 to 8 weeks; ② High-dose dexamethasone: oral or intravenous dose of 40 mg per day, continued for 4 days, and then stopped. If the patient's platelet count is persistently below 30×10 9 / L or if bleeding symptoms persist within 10 days after medication, another 4-day course of dexamethasone (40 mg per day) should be given.

[0088] The platelet count of the subjects was tested within 1 month after the administration of glucocorticoids (the day of administration was recorded as the first day of testing). The platelet count was tested twice in a row with an interval of at least 7 days within 1 month, and the judgment was made as follows:

[0089] Drug reaction group (R): Both tests meet the following conditions: increased platelet count (≥30×10 9 / L), and the platelet count increased by at least 2 times the baseline platelet count, and was not accompanied by bleeding; that is, there was a drug response.

[0090] Drug non-response group (NR): at least one of the two examinations meets the following conditions: platelet count <30×10 9 / L or platelet count increase < 2-fold over baseline platelet count or bleeding, i.e. no drug response.

[0091] Example 1: Construction of an early drug response model for glucocorticoid treatment in ITP patients

[0092] 1. Obtaining stool samples from subjects

[0093] 100 ITP patients (meeting the inclusion criteria) were used as the training set. These subjects kept stool samples before receiving standard glucocorticoid treatment. Then standard glucocorticoid treatment was performed, and the platelet count of the subjects was detected within 1 month after glucocorticoid treatment according to the above method. The subjects were divided into R group (71 cases) and NR group (29 cases).

[0094] Informed consent was obtained from all subjects in this trial.

[0095] 2. Screening of microbial markers

[0096] Fecal samples of 100 ITP patients in the training set were collected, and DNA was extracted from each sample (200 mg) using the QIAamp stool DNA extraction kit (Qiagen, Germany). Metagenomic sequencing was performed to obtain the relative abundance values ​​of species abundance annotations in each sample. The microbiome data contains multiple classification levels, including phylum, class, order, family, genus, and species. The Mann-Whitney test was performed based on the relative abundance table, and the top 10 species-level difference classification data sets with the most significant p values ​​between the R group and the NR group were obtained (screened by average relative abundance greater than 0.01%) (Table 1 and Figure 2 As shown in the figure above), as effective microorganisms.

[0097] Table 1 shows the first 10 species-level difference classification datasets

[0098]

[0099]

[0100] In the above table, the P value is the result of the Mann-Whitney test between the R group and the NR group, and log2FC represents the logarithm of the ratio of the relative abundance of each microorganism in the R group to the relative abundance of the NR group with base 2.

[0101] 3. Build a model through training set

[0102] Figure 1 Flowchart of the model for predicting the response status of glucocorticoid treatment in patients with ITP.

[0103] 1. Obtaining effective microbial abundance

[0104] The relative abundances (%) of the top 10 effective microorganisms in the stool samples of 100 ITP patients in the above two training sets were tabulated, and the results are shown in Table 2.

[0105] Table 2 shows the relative abundance of the top 10 effective microorganisms

[0106]

[0107]

[0108]

[0109]

[0110] In the above table, A is Bacteroides xylanisolvens, B is Bacteroides sp.3_1_23, C is Bacteroides ovatus, D is Bacteroides sp.D2, E is Ruminococcus faecis, F is Parabacteroides gordonii, G is Bacteroides sp.2_2_4, H is Turicibacter sanguinis, I is Desulfovibrio sp.An276, and J is Megamonas rupellensis.

[0111] 2. Calculate the alpha diversity index

[0112] The metagenomic sequencing results of the stool samples of the above 100 ITP patients were calculated using the R language vegan package (refer to Oksanen, Jari, et al. "vegan: Community Ecology Package." R package, version 2.0-10, 2013. https: / / CRAN.R-project.org / package=vegan.) The alpha diversity index of each sample was calculated based on the species relative abundance table. During the model development process, the Chao1 index, Dominance index, and Shannon index were used in combination with all metagenomic sequencing data to evaluate the diversity of the microbiome at each taxonomic level.

[0113] Alpha diversity indicators include Chao1 index, Dominance index and Shannon index.

[0114] The Chao1 index is a nonparametric statistical method used to estimate the number of unobserved species (species richness) in a community. The larger the value, the higher the species richness in the community. Its calculation formula is shown in formula 1:

[0115] S Chao1 =S obs +2F1+2F2 (Formula 1);

[0116] Among them, S obsis the number of all species at the species level in the metagenomic sequencing results of the sample, F1 is the number of species that appear once at the species level in the metagenomic sequencing results of the sample, and F2 is the number of species that appear twice at the species level in the metagenomic sequencing results of the sample.

[0117] Dominance index is an indicator to measure the degree of dominance of a species or a few species in a community (species uniformity). The closer its value is to 1, the more dominant a species or a few species are. Its calculation formula is shown in formula 2:

[0118]

[0119] Among them, S obs is the number of all species at the species level in the metagenomic sequencing results of the sample, p i It is the ratio of the number of individuals of the i-th species level to the total number of individuals of all species at the species level.

[0120] The Shannon index is an index to measure community diversity. It takes into account species richness and uniformity. It is also called the species diversity index or information entropy index. The larger the value, the higher the diversity of the community. Its calculation formula is shown in Formula 3:

[0121]

[0122] Among them, S obs is the number of all species at the species level in the metagenomic sequencing results of the sample, p i It is the ratio of the number of individuals of the i-th species level to the total number of individuals of all species at the species level.

[0123] The species-level alpha diversity indicators Chao1 index, Dominance index, and Shannon index were selected to be included in the model construction to reflect the overall characteristics of the microbiome. The results of the three alpha diversity indicators of the fecal microbiome of 100 ITP patients in the training set are shown in Tables 3 and Figure 2 Picture below.

[0124] Table 3 shows the three alpha diversity indicators of the fecal microbiome

[0125]

[0126]

[0127]

[0128]

[0129]

[0130]

[0131] 3. Obtain clinical indicators

[0132] According to the above implementation method, clinical indicators of 100 ITP patients in the training set were collected, including gender (0 represents female, 1 represents male), age (in years), course of disease (time from the first diagnosis of ITP to the time before the current glucocorticoid treatment (treatment after collecting stool samples), in months), platelet count before glucocorticoid treatment (in 10 9 / L), World Health Organization bleeding score before glucocorticoid treatment, and whether the patient had received glucocorticoid treatment in the past (0 represents no, 1 represents yes). The results are shown in Table 4.

[0133] Table 4 shows the clinical indicators

[0134]

[0135]

[0136]

[0137]

[0138]

[0139]

[0140] 4. Model construction

[0141] The relative abundance of 10 microorganisms, 3 alpha diversity indices, 6 clinical indicators and the known glucocorticoid treatment drug response effects (R and NR distinguished by platelet count detection) of each sample obtained in 1, 2, and 3 above were used as input data for modeling. Random forest (RF) and gradient boosting machine (GBM) calculation methods were used for training, respectively, to construct random forest models and gradient boosting machine models, respectively, and output the predicted probability values ​​of glucocorticoid treatment drug response (Table 5).

[0142] Random forest model: This study used the random forest algorithm to train the input data and output the predicted probability value of glucocorticoid treatment drug response. The optimization parameters of the model were determined by grid search, and the key parameters finally selected included: the number of decision trees (nesimators) was 150, the maximum tree depth (max_depth) was 10, the minimum number of split node samples (min_samples_split) was 5, the minimum number of leaf node samples (min_samples_leaf) was 2, and the split criterion was calculated based on the Gini coefficient (criterion = gini). The model was trained using ten repeated ten-fold cross validation to ensure the robustness and generalization ability of the model. Based on the predicted probability value of glucocorticoid treatment drug response for each sample in the training set, the receiver operating characteristic (ROC) curve was drawn, and the AUC (area under the curve) value was calculated. At the same time, the optimal threshold was determined to be 0.74 in combination with the Precision and Recall curves.

[0143] If the probability value is greater than the set threshold, the subject is predicted to be or is a candidate for glucocorticoid treatment drug response (R').

[0144] If the probability value is less than or equal to the set threshold, the subject is predicted to be or is a candidate for glucocorticoid treatment drug non-responder (NR').

[0145] Gradient boosting machine model: The hyperparameter optimization of the gradient boosting machine model is also achieved through grid search, which trains the input data and outputs the predicted probability value of glucocorticoid treatment drug response. The parameters finally selected for this model include: the number of decision trees (n_estimators) is 200, the learning rate (learning_rate) is 0.05, the maximum tree depth (max_depth) is 8, the minimum number of split node samples (min_samples_split) is 4, and the minimum number of leaf node samples (min_samples_leaf) is 3. The model also uses ten repeated ten-fold cross-validation to improve robustness, and the optimal threshold is 0.976230929. Based on the set threshold, the following prediction rules are generated:

[0146] If the probability value is greater than the set threshold, the subject is predicted to be or is a candidate for glucocorticoid treatment drug response (R').

[0147] If the probability value is less than or equal to the set threshold, the subject is predicted to be or is a candidate for glucocorticoid treatment drug non-responder (NR').

[0148] Table 5 shows the prediction results of glucocorticoid treatment drug response for the random forest model and gradient boosting model training set samples.

[0149]

[0150]

[0151]

[0152] In the above table, Actual Group represents the platelet count detection area grouping, 0 represents no response to glucocorticoid therapy (NR), and 1 represents a response to glucocorticoid therapy (R); Predicted Probability is the predicted probability value, Predicted Group represents the RF model or GBM model prediction grouping, 0 represents no response to glucocorticoid therapy (NR'), and 1 represents a response to glucocorticoid therapy (R').

[0153] 5. Verification of model ROC curve

[0154] The model was built using the training set, and the hyperparameters were optimized using grid search and cross-validation techniques. The training set was used as the internal validation set, and 10-fold cross-validation was performed to evaluate the robustness and reliability of the model. Specifically:

[0155] The accuracy, specificity, and sensitivity of the model were further evaluated by the confusion matrix. The confusion matrix divides the true labels and model prediction results into four categories: True Positive (TP), False Positive (FP), True Negative (TN), and False Negative (FN). The accuracy, specificity, and sensitivity were calculated based on the results of the confusion matrix. In the training set: the accuracy of the random forest model was 95.0%, the specificity was 93.0%, and the sensitivity was 97.0%. The accuracy of the gradient boosting machine model was 94.0%, the specificity was 92.0%, and the sensitivity was 96.0%.

[0156] ROC curve was drawn using R software.

[0157] The results are as follows Figure 3 As shown in the figure, it can be seen that the constructed RF model and GBM model can effectively distinguish the samples in the R group and the NR group, and the area under the ROC curve of both is 1.00. Both models can be used to predict the drug response effect of glucocorticoid treatment of ITP in subjects.

[0158] Example 2: Validating the model using a validation set

[0159] 1. Obtaining a stool sample

[0160] Another 52 ITP patients (who met the inclusion criteria and were sampled before standard glucocorticoid treatment) were used as a validation set. These subjects kept stool samples before receiving standard glucocorticoid treatment and then received standard glucocorticoid treatment.

[0161] 2. Obtaining microbial markers, clinical indicators and drug response effects for each sample

[0162] 1. Effective microbial abundance

[0163] Fecal samples for the validation set were collected, and DNA was extracted and library was constructed, followed by metagenomic sequencing to detect the relative abundance of each microorganism in the feces.

[0164] The relative abundances of these 10 microorganisms, Bacteroides xylanisolvens, Bacteroides sp.3_1_23, Bacteroides ovatus, Bacteroides sp.D2, Ruminococcus faecis, Parabacteroides gordonii, Bacteroides sp.2_2_4, Turicibacter sanguinis, Desulfovibrio sp.An276, and Megamonasrupellensis, were recorded as effective microbial abundance (Table 6).

[0165] 2. Calculation of alpha diversity index

[0166] According to the method of Example 1, three alpha diversity indices of each sample were obtained: Chao1 index, Dominance index and Shannon index (Table 6).

[0167] Table 6 shows the relative abundance of 10 microorganisms and three alpha diversity indicators

[0168]

[0169]

[0170] In the above table, A is Bacteroides xylanisolvens, B is Bacteroides sp.3_1_23, C is Bacteroides ovatus, D is Bacteroides sp.D2, E is Ruminococcus faecis, F is Parabacteroides gordonii, G is Bacteroides sp.2_2_4, H is Turicibacter sanguinis, I is Desulfovibrio sp.An276, and J is Megamonas rupellensis.

[0171] 3. Clinical indicators and drug response effects

[0172] According to the aforementioned implementation, clinical indicators of 52 ITP patients in the validation set were collected, and the results are shown in Table 7. At the same time, the platelet count of the subjects was tested within 1 month after the administration of glucocorticoids and evaluated using R and NR (Table 7).

[0173] Table 7 shows clinical indicators and drug response effects

[0174]

[0175]

[0176] 3. Verify the model

[0177] The relative abundance of 10 microorganisms, 3 alpha diversity indices, and 6 clinical indicators of each sample obtained in steps 1, 2, and 3 of step 2 were used as input data and respectively input into the two prediction models (random forest model and gradient boosting model) established in step 4 of step 3 of Example 1. The model outputs a predicted probability value. Based on the comparison between the predicted probability value and the threshold, the drug response effect of glucocorticoid treatment in ITP subjects is predicted (Table 8).

[0178] Table 8 shows the prediction results of glucocorticoid treatment drug response of the random forest model and gradient boosting model test set samples

[0179]

[0180]

[0181]

[0182] In the above table, Actual Group represents the platelet count detection area grouping, 0 represents no response to glucocorticoid therapy (NR), and 1 represents a response to glucocorticoid therapy (R); Predicted Probability is the predicted probability value, Predicted Group represents the RF model or GBM model prediction grouping, 0 represents no response to glucocorticoid therapy (NR'), and 1 represents a response to glucocorticoid therapy (R').

[0183] The prediction results are evaluated based on the actual results (actually NR or R).

[0184] Results Figure 4 , the gradient boosting model has the best effect, with an AUC value of 0.76.

[0185] The accuracy, specificity, and sensitivity of the model are further evaluated through the confusion matrix. The confusion matrix divides the true labels and model prediction results into four categories: true positive (TP), false positive (FP), true negative (TN), and false negative (FN). The accuracy, specificity, and sensitivity are calculated based on the results of the confusion matrix.

[0186] In the random forest model:

[0187] Training set: AUC = 1.00, accuracy = 100.0%, sensitivity = 100.0%.

[0188] Validation set: AUC = 0.73, accuracy = 90.4%, sensitivity = 100.0%.

[0189] Gradient boosting machine model:

[0190] Training set: AUC = 1.00, accuracy = 100.0%, sensitivity = 100.0%.

[0191] Validation set: AUC = 0.76, accuracy = 90.4%, sensitivity = 100.0%.

[0192] The present invention has been described in detail above. It will be apparent to those skilled in the art that the present invention may be implemented in a wide range under equivalent parameters, concentrations and conditions without departing from the spirit and scope of the present invention and without the need for unnecessary experimentation. Although the present invention provides specific embodiments, it should be understood that further improvements may be made to the present invention. In short, according to the principles of the present invention, this application intends to include any changes, uses or improvements to the present invention, including changes made by conventional techniques known in the art that depart from the scope disclosed in this application. Applications of some of the basic features may be made within the scope of the following appended claims.

Claims

1. A computer device comprising a memory, a processor and a computer program stored in the memory, characterized in that: The processor executes the computer program to implement the following steps: S1. Receive data: Receive sample data, which includes 10 relative abundances of microorganisms at the species level, 3 alpha diversity indices, and 6 clinical indicators in the ex vivo fecal samples of the subjects; The 10 species-level microorganisms are as follows: Bacteroides xylanisolvens, Bacteroides sp.3_1_23, Bacteroides ovatus, Bacteroides sp.D2, Ruminococcus faecis, Parabacteroidesgordonii, Bacteroides sp.2_2_4, Turicibacter sanguinis, Desulfovibrio sp.An276 and Megamonas rupellensis; The three alpha diversity indices are Chao1 index, Dominance index and Shannon index; The six clinical indicators include gender, age, disease duration, platelet count before glucocorticoid treatment, World Health Organization bleeding score before glucocorticoid treatment, and whether glucocorticoid treatment was received in the past; S2. Input data: input the sample data into the prediction model of glucocorticoid treatment drug response effect in ITP patients; The glucocorticoid treatment drug response effect prediction model for ITP patients is constructed according to a method comprising the following steps: using the following data in in vitro fecal samples of ITP patients with known glucocorticoid treatment drug response and ITP patients with known glucocorticoid treatment drug non-response and the drug response or drug non-response corresponding to each sample as training samples to train the glucocorticoid treatment drug response effect model for ITP patients, wherein the following data are: the relative abundance of the 10 species-level microorganisms, the 3 alpha diversity indices and the 6 clinical indicators; The glucocorticoid treatment drug response effect of the ITP patient is drug response or drug non-response; S3, outputting the result: outputting the probability value of the predicted glucocorticoid treatment drug response effect of the subject through the glucocorticoid treatment drug response effect model of ITP patients; and then calculating the glucocorticoid treatment drug response effect of the subject based on the probability value; The subject is an ITP patient.

2. A computer program product, comprising a computer program, characterized in that: When the computer program is executed by a processor, the steps in the computer device of claim 1 are implemented.

3. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and the computer program causes a computer to execute the steps in the computer device according to claim 1.

4. A device for predicting the effect of glucocorticoid drug response in the treatment of ITP, the device comprising: S1, data receiving module: used to receive sample data, wherein the sample data is the relative abundance of 10 species-level microorganisms, 3 alpha diversity indices and 6 clinical indicators in the ex vivo fecal samples of the subjects; The 10 species-level microorganisms are as follows: Bacteroides xylanisolvens, Bacteroides sp.3_1_23, Bacteroides ovatus, Bacteroides sp.D2, Ruminococcus faecis, Parabacteroidesgordonii, Bacteroides sp.2_2_4, Turicibacter sanguinis, Desulfovibrio sp.An276 and Megamonas rupellensis; The three alpha diversity indices are Chao1 index, Dominance index and Shannon index; The six clinical indicators include gender, age, disease duration, platelet count before glucocorticoid treatment, World Health Organization bleeding score before glucocorticoid treatment, and whether glucocorticoid treatment was received in the past; S2, data input module: used to input the sample data into the prediction model of glucocorticoid treatment drug response effect in ITP patients; The glucocorticoid treatment drug response effect prediction model for ITP patients is constructed according to a method comprising the following steps: using the following data in in vitro fecal samples of ITP patients with known glucocorticoid treatment drug response and ITP patients with known glucocorticoid treatment drug non-response and the drug response or drug non-response corresponding to each sample as training samples to train the glucocorticoid treatment drug response effect model for ITP patients, wherein the following data are: the relative abundance of the 10 species-level microorganisms, the 3 alpha diversity indices and the 6 clinical indicators; The glucocorticoid treatment drug response effect of the ITP patient is drug response or drug non-response; S3, result output module: used for outputting a probability value of predicting the glucocorticoid treatment drug response effect of the subject through the glucocorticoid treatment drug response effect model of the ITP patient; and then calculating the glucocorticoid treatment drug response effect of the subject based on the probability value; The subject is an ITP patient.

5. A method for predicting or assisting in predicting the drug response effect of glucocorticoids in treating ITP in a subject, characterized in that: The method comprises the following steps: S1. Obtain the relative abundance of 10 species-level microorganisms, 3 alpha diversity indices and 6 clinical indicators in the subjects' in vitro fecal samples; The 10 species-level microorganisms are as follows: Bacteroides xylanisolvens, Bacteroides sp.3_1_23, Bacteroides ovatus, Bacteroides sp.D2, Ruminococcus faecis, Parabacteroidesgordonii, Bacteroides sp.2_2_4, Turicibacter sanguinis, Desulfovibrio sp.An276 and Megamonas rupellensis; The three alpha diversity indices are Chao1 index, Dominance index and Shannon index; The six clinical indicators include gender, age, disease duration, platelet count before glucocorticoid treatment, World Health Organization bleeding score before glucocorticoid treatment, and whether glucocorticoid treatment was received in the past; S2, inputting the sample data into a prediction model for the effect of glucocorticoid treatment on ITP patients; The glucocorticoid treatment drug response effect prediction model for ITP patients is constructed according to a method comprising the following steps: using the following data in in vitro fecal samples of ITP patients with known glucocorticoid treatment drug response and ITP patients with known glucocorticoid treatment drug non-response and the drug response or drug non-response corresponding to each sample as training samples to train the glucocorticoid treatment drug response effect model for ITP patients, wherein the following data are: the relative abundance of the 10 species-level microorganisms, the 3 alpha diversity indices and the 6 clinical indicators; The glucocorticoid treatment drug response effect of the ITP patient is drug response or drug non-response; S3, predicting the probability value of the subject's predicted glucocorticoid treatment drug response effect through the output of the glucocorticoid treatment drug response effect model for ITP patients; and then calculating the glucocorticoid treatment drug response effect of the subject based on the probability value; The subject is an ITP patient.

6. A method for constructing a prediction model for the effect of glucocorticoid treatment on ITP patients, characterized in that: The method comprises the following steps: using the following data in in vitro fecal samples of ITP patients with known glucocorticoid treatment drug response subjects and ITP patients with known glucocorticoid treatment drug non-response subjects and the drug response or drug non-response corresponding to each sample as training samples to train an ITP patient glucocorticoid treatment drug response effect model, wherein the following data are: the relative abundance of the 10 species-level microorganisms, the 3 alpha diversity indices and the 6 clinical indicators; The glucocorticoid treatment drug response effect of the ITP patient is drug response or drug non-response.

Images