TIGIT target-based sepsis risk prediction method and system

Abstract

The invention relates to the technical field of sepsis prediction, and discloses a sepsis risk prediction method and system based on a TIGIT target spot. The prediction method comprises the following steps: firstly, obtaining a sepsis initial disease probability through a sepsis first prediction model; determining a method for predicting the sepsis illness probability according to the sepsis initial illness probability and the TIGIT; if the probability is not higher than the preset probability threshold value, determining that the probability is low, and inputting the sample data with the low disease probability into a sepsis low-probability prediction model to obtain the sepsis disease probability; and finally, obtaining a final sepsis prediction result according to the initial sepsis disease probability and the sepsis disease probability. By maximizing the effect of the TIGIT in the prediction process, the precision of sepsis prediction is improved on the whole.

Description

Technical Field

[0001] The present invention relates to the technical field of sepsis prediction, and in particular to a method and system for predicting sepsis risk based on TIGIT targets. Background Art

[0002] Sepsis is a life-threatening organ dysfunction caused by a dysregulated response to infection. Its severity can be roughly divided into sepsis, severe sepsis, and septic shock. The causes of sepsis are usually related to various types of trauma and infection, such as pneumonia, peritonitis, and urinary tract infection. When the body is infected, bacteria, toxins and other components enter the body, activate inflammatory cells, produce a large number of inflammatory mediators, and cause local or systemic inflammatory reactions. If the inflammatory response intensifies and causes deep suppression of immune function, sepsis may occur.

[0003] To this day, sepsis remains one of the most common causes of death in critically ill patients, so the prediction of sepsis is crucial for early identification and timely treatment. Accurate prediction can help doctors take intervention measures before the condition worsens, thereby reducing the patient's mortality rate and risk of complications. In recent years, with the advancement of medical research, large-sample, multi-center clinical randomized controlled studies have provided more evidence-based medicine for the treatment of sepsis. In addition, with the rapid development of big data and artificial intelligence technology, there has also been a major breakthrough in the prediction and diagnosis of sepsis.

[0004] However, there are many types of biomarkers for diagnosing sepsis, among which the expression level of TIGIT is positively correlated with the severity of sepsis or septic shock, and is the gold standard for diagnosing sepsis. If TIGIT is used as the only indicator for diagnosing sepsis, the diagnostic deviation will be too large. If it is used together with other biomarkers for the prediction of sepsis, it will face the problem of how to set the weight relationship between TIGIT and other biomarkers to maximize its important role in the prediction of sepsis, which reduces the accuracy of the prediction.

[0005] In the prior art, an initial prediction probability may be used as a basis for determining which classification model to use for the next step of accurate prediction. However, using an inaccurate initial prediction probability as a basis for classification judgment will further reduce the accuracy of the final prediction result.

[0006] Therefore, there is an urgent need for a sepsis risk prediction method and system based on the TIGIT target to improve the prediction accuracy of sepsis. Summary of the invention

[0007] In order to solve the above technical problems, the present invention provides a method for predicting sepsis risk based on TIGIT targets, and the prediction method comprises the following steps: Obtaining initial sample data for predicting sepsis; Inputting the initial sample data into a first sepsis prediction model to obtain an initial probability of sepsis; The method for predicting the probability of sepsis is determined by comparing the initial probability of sepsis with a preset probability threshold, including the following situations: If the initial probability of sepsis is higher than the preset probability threshold, and TIGIT is greater than the content threshold, it is a high probability of illness, and the high probability of illness sample data is input into the sepsis high probability prediction model to obtain the probability of sepsis; In other cases, the probability of illness is low, and the sample data of low probability of illness are input into the low probability prediction model of sepsis to obtain the probability of sepsis. According to the initial probability of sepsis and the probability of sepsis, the final sepsis prediction result was determined by weight analysis method.

[0008] Furthermore, the initial sample data for predicting sepsis include C-reactive protein, procalcitonin, lactic acid, white blood cell count, platelet count, sequential organ failure score, interleukin 6, interleukin 3, endotoxin, exotoxin, prothrombin time, thromboplastin time, D-dimer, blood creatinine, urine volume, total bilirubin, alanine aminotransferase, and aspartate aminotransferase.

[0009] Furthermore, the high probability of disease sample data includes C-reactive protein, procalcitonin, lactic acid, white blood cell count, platelet count, sequential organ failure score, interleukin 6, interleukin 3, and TIGIT.

[0010] Furthermore, the low probability of disease sample data includes TIGIT, C-reactive protein, endotoxin, exotoxin, prothrombin time, thromboplastin time, D-dimer, blood creatinine, urine volume, total bilirubin, alanine aminotransferase, and aspartate aminotransferase.

[0011] Furthermore, the method for training the sepsis low probability prediction model is: Acquire historical data of low disease probability samples, wherein the historical data includes multiple groups of low disease probability sample data and the sepsis disease probability corresponding to each group of low disease probability sample data; The low probability sepsis prediction model is trained using the low probability sample historical data until a training end condition is met.

[0012] Further, using the low probability sample historical data to train a sepsis low probability prediction model comprises the following steps: Obtaining a training data set by standardizing the historical data of the low disease probability samples; Using the training data set to train a low probability sepsis prediction model until a corresponding training end condition is met, thereby obtaining a low probability sepsis initial prediction model; Calculating the contribution of each biomarker in the training data set to the low probability initial prediction model of sepsis according to the characteristic gradient calculation method; Based on the feature importance evaluation rule, the contribution is evaluated and the training data set is reconstructed; The reconstructed training data set is used to train the initial prediction model for the low probability of sepsis until the corresponding training end conditions are met, thereby obtaining the final prediction model for the low probability of sepsis.

[0013] Furthermore, based on the feature importance evaluation rule, the contribution is evaluated and the training data set is reconstructed as follows: Arrange the contribution degrees in descending order; The training data of biomarkers corresponding to the top N contributions are selected to reconstruct the training data set.

[0014] Furthermore, the calculation formula for the contribution of each biomarker to the low probability initial prediction model of sepsis is: ; In the formula, Ψ j represents the contribution of the j-th biomarker, S represents the feature subset excluding the j-th biomarker, |S| represents the number of elements in the set S, |N| represents the total number of biomarkers, f(S) represents the prediction result of the low probability initial prediction model for sepsis on the feature subset S, and f(S∪{j}) represents the prediction result of the low probability initial prediction model for sepsis after adding the feature subset of the j-th biomarker to the feature subset S.

[0015] Another aspect of the present invention further provides a sepsis risk prediction system based on TIGIT target, which is used to execute any of the above-mentioned sepsis risk prediction methods based on TIGIT target, and the prediction system comprises: An initial sample data acquisition module, used to acquire initial sample data for predicting sepsis; a sepsis initial probability prediction module, connected to the initial sample data acquisition module, and obtaining the initial probability of sepsis by inputting the initial sample data into the first sepsis prediction model; A sepsis probability prediction module is connected to the sepsis initial probability prediction module and is used to determine a method for predicting the sepsis probability based on the comparison between the sepsis initial probability and a preset probability threshold, including the following situations: If the initial probability of sepsis is higher than the preset probability threshold, and TIGIT is greater than the content threshold, it is a high probability of illness, and the high probability of illness sample data is input into the sepsis high probability prediction model to obtain the probability of sepsis; In other cases, the probability of illness is low, and the sample data of low probability of illness are input into the low probability prediction model of sepsis to obtain the probability of sepsis. The output module is connected to the sepsis probability prediction module and is used to determine the final sepsis prediction result through a weight analysis method according to the initial sepsis probability and the sepsis probability.

[0016] The embodiments of the present invention have the following technical effects: The sepsis risk prediction method based on TIGIT target provided in this application divides the prediction process into three stages: In the first stage, we first obtained a rough estimate of the initial probability of sepsis without using TIGIT as the initial sample data, to avoid the problem of other biomarkers weakening their importance in the diagnostic process; In the second stage, the sepsis probability is obtained by using the high / low probability prediction model of sepsis based on the initial probability of sepsis and the TIGIT content. This not only fully reflects the importance of TIGIT in diagnosing sepsis, but also overcomes the inaccuracy of using only the inaccurate initial probability of sepsis as the basis for judgment, further improving the accuracy of the final prediction results. In the third stage, the final sepsis prediction result is determined by weight analysis based on the initial probability of sepsis and the sepsis probability. The final prediction result not only takes into account the impact of different types of sample data on the prediction result, but also takes into account the more targeted sepsis probability obtained by different prediction models for sample data with high and low prevalence probabilities, thereby improving the overall prediction result of sepsis. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0018] Figure 1 is a flow chart of a method for predicting sepsis risk based on TIGIT targets provided by an embodiment of the present invention; Figure 2 is a graph showing the expression of TIGIT content in mouse spleen tissue in an embodiment of the present invention; Figure 3 It is the MFI expression diagram of the experimental group and the control group in the embodiment of the present invention; Figure 4 It is a survival rate comparison chart of the TIGIT knockout experimental results in the embodiments of the present invention. DETAILED DESCRIPTION

[0019] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be described clearly and completely below. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work belong to the scope of protection of the present invention.

[0020] Based on the problem that the accuracy of sepsis prediction results is not high, the present invention provides a sepsis risk prediction method based on TIGIT targets. Figure 1 It can be seen that the prediction method includes the following steps: S1: Obtain initial sample data for predicting sepsis; Furthermore, the initial sample data for predicting sepsis include C-reactive protein, procalcitonin, lactic acid, white blood cell count, platelet count, sequential organ failure score, interleukin 6, interleukin 3, endotoxin, exotoxin, prothrombin time, thromboplastin time, D-dimer, blood creatinine, urine volume, total bilirubin, alanine aminotransferase, and aspartate aminotransferase.

[0021] The expression level of TIGIT is positively correlated with the severity of sepsis or septic shock, and it is the gold standard for diagnosing sepsis. However, when predicting sepsis together with other biomarkers, the impact of various biomarkers on the corresponding sepsis prediction results needs to be taken into account during model learning, which will inevitably weaken the importance of TIGIT. Therefore, this application first obtains a rough estimate of the initial probability of sepsis without using TIGIT as the initial sample data, avoiding the problem of other biomarkers weakening their importance in the diagnosis process, but instead applies it to the subsequent judgment of the basis for using the high / low probability prediction model of sepsis.

[0022] C-reactive protein is an acute phase protein whose plasma concentration increases during sepsis. Procalcitonin (PCT) is produced by the thyroid gland and neuroendocrine cells of the lungs and intestines during bacterial infection and is used to diagnose bacterial infection and assess the severity of sepsis. Increased lactate levels may indicate tissue hypoxia, which is common in septic shock and multiple organ dysfunction. White blood cell counts may increase during sepsis, but may also be normal or decreased. Platelet counts may decrease in patients with sepsis. The Sequential Organ Failure Assessment (SOFA) is a tool to assess the degree of organ dysfunction and is used to diagnose sepsis. Inflammatory markers such as interleukin 6 and interleukin 3 increase during sepsis. Endotoxins and exotoxins may be detected during sepsis. Patients with sepsis may have coagulation disorders, so prothrombin time, thromboplastin time, and D-dimers may also change. Like serum creatinine and urine volume, sepsis may also cause acute kidney injury, and both serum creatinine and urine volume will change after sepsis. At the same time, sepsis can also affect liver function. Total bilirubin, alanine aminotransferase, and aspartate aminotransferase are also important biomarkers for diagnosing sepsis.

[0023] S2: inputting the initial sample data into a first sepsis prediction model to obtain an initial probability of sepsis; S3: Determine a method for predicting the probability of sepsis based on the comparison of the initial probability of sepsis with a preset probability threshold.

[0024] In the prior art, when predicting a disease, an initial prediction probability may be used as a basis for determining which classification model to use for the next step of accurate prediction. However, using an inaccurate initial prediction probability as a basis for classification judgment will further reduce the accuracy of the final prediction result.

[0025] When predicting sepsis, the present application adds TIGIT, and jointly determines the sepsis probability using the high / low probability sepsis prediction model together with the initial sepsis probability to obtain the sepsis probability. This can not only fully utilize the importance of TIGIT in diagnosing sepsis, but also overcome the inaccuracy of using only the inaccurate initial sepsis probability as the basis for judgment, thereby further improving the accuracy of the final prediction results.

[0026] The preset probability threshold can be set in combination with the distribution of the initial sample data: if there are a large number of low-risk samples in the initial sample data set, but fewer high-risk samples, the probability threshold can be increased to help the model focus on identifying patients who are truly at high risk. This can improve the specificity of the model and reduce the number of low-risk patients misclassified as high-risk. If there are more high-risk samples and fewer low-risk samples in the initial sample data set, lowering the threshold can help the model capture more high-risk patients. This can increase the sensitivity of the model and ensure that more truly high-risk patients are identified.

[0027] S3A: If the initial probability of sepsis is higher than the preset probability threshold, and TIGIT is greater than the content threshold, it is a high probability of sepsis, and the high probability sample data is input into the high probability sepsis prediction model to obtain the probability of sepsis; TIGIT (T cell immunoglobulin and ITIM domain protein) is a co-inhibitory receptor expressed in regulatory T cells, activated CD8+ and CD4+ T cells, and natural killer (NK) cells. In their studies, the inventors found that the expression level of TIGIT on peripheral blood CD4+ T cells in patients with sepsis or septic shock was significantly higher than that in healthy volunteers, and was positively correlated with the severity of sepsis or septic shock, and could be used as a gold indicator for the diagnosis of sepsis.

[0028] Abnormal Tregs (regulatory T cells) is a key link in the disorder of immune response in sepsis. The expression of Tregs co-inhibitory receptor TIGIT is upregulated in sepsis patients, which is directly related to its activity and disease severity. TIGIT regulates Tregs activity, affects macrophage M2 polarization, and participates in the occurrence and development of immune disorders in sepsis.

[0029] The inventors of the present application confirmed the effects of changes in Treg number and activity on macrophage polarization and sepsis through antibody blocking, Treg adoptive transfer, etc.; using in vitro co-culture and Treg conditional TIGIT knockout mice, it was confirmed that Tregs activity is regulated by TIGIT expression, and the molecular mechanism affecting macrophage polarization also confirmed that TIGIT knockout can improve immune disorders such as Tregs changes and macrophage M2 polarization in septic mice.

[0030] Figure 2 is a graph showing the expression of TIGIT content in mouse spleen tissue in an embodiment of the present invention, Figure 3 The MFI expression diagrams of the experimental group and the control group in the embodiment of the present invention are used to show the expression of TIGIT in the spleen tissue of mice subjected to cecum ligation and puncture (CLP). The expression level of TIGIT in the spleen of CLP mice was significantly increased compared with that of the sham injury group. Figure 2 The middle histogram represents the percentage of TIGIT expression in spleen. Figure 3 The middle bar graph represents the fluorescence intensity of TIGIT expression in the experimental group and the control group. There were 6 mice in each group, and the values ​​were expressed as mean ± SD variance. ( Compared with the sham injury group, P<0.05).

[0031] Figure 4 This is a survival comparison chart of the TIGIT knockout experiment results in the embodiment of the present invention. Compared with wild-type (TIGIT knockout group) mice, the survival rate of CLP mice (TIGIT non-knockout group) is significantly increased. This confirms that the expression level of TIGIT after sepsis is directly related to the severity of the disease, and the expression level of TIGIT can affect the survival rate and inflammatory response intensity of sepsis.

[0032] Furthermore, the high probability sample data include C-reactive protein, procalcitonin, lactic acid, white blood cell count, platelet count, sequential organ failure score, interleukin 6, interleukin 3, and TIGIT. The expression levels of these biomarkers are usually significantly increased in the early or severe stages of sepsis, and therefore have high sensitivity and specificity.

[0033] S3B: In other cases, the probability of illness is low, and the sample data of low probability of illness are input into the low probability prediction model of sepsis to obtain the probability of sepsis. The initial probability of sepsis is obtained by using a larger variety of initial sample data through the first sepsis prediction model, and the sepsis probability is obtained by using a more targeted high / low probability sample data and a more accurate sepsis high / low probability prediction model. The high / low probability sample data is highly targeted, has consistent distribution characteristics, and has fewer types, which can improve the prediction accuracy of the sepsis high / low probability prediction model and improve efficiency.

[0034] In detail, using the prediction results of the first sepsis prediction model, the initial sample data is divided into two major situations, corresponding to high probability of illness sample data and low probability of illness sample data, reducing the amount of data for the high probability sepsis prediction model and the low probability sepsis prediction model, and being more targeted, which is conducive to model learning, making it easier to train a model with better performance, improving the accuracy of the prediction results of the high probability sepsis prediction model and the low probability sepsis prediction model, and improving the computational efficiency.

[0035] Furthermore, the low probability of illness sample data includes TIGIT, C-reactive protein, endotoxin, exotoxin, prothrombin time, thromboplastin time, D-dimer, blood creatinine, urine volume, total bilirubin, alanine aminotransferase, aspartate aminotransferase. In the early stage of sepsis or mild infection, the effects on coagulation function, renal function, and liver function are not obvious, so the corresponding biomarkers change slowly or more steadily, and the expression level is not obvious. C-reactive protein changes significantly during the inflammatory process. When other indicators of inflammation may not show obvious abnormalities in the early stage, it may still be abnormally expressed, so it is also included in the low probability of illness sample data.

[0036] Because both the high probability sample data and the low probability sample data have been screened in the initial sample data, the number of sample types has been reduced, and the disadvantage of TIGIT being reduced in importance by other markers has been overcome to a certain extent. At the same time, adding the sample data of the low probability sepsis prediction model and the high probability sepsis prediction model to TIGIT can improve the accuracy of the prediction.

[0037] Furthermore, the method for training the sepsis low probability prediction model is: Acquire historical data of low disease probability samples, wherein the historical data includes multiple groups of low disease probability sample data and the sepsis disease probability corresponding to each group of low disease probability sample data; The low probability sepsis prediction model is trained using the low probability sample historical data until a training end condition is met.

[0038] Further, using the low probability sample historical data to train a sepsis low probability prediction model comprises the following steps: Obtaining a training data set by standardizing the historical data of the low disease probability samples; Using the training data set to train a low probability sepsis prediction model until a corresponding training end condition is met, thereby obtaining a low probability sepsis initial prediction model; Calculating the contribution of each biomarker in the training data set to the low probability initial prediction model of sepsis according to the characteristic gradient calculation method; Based on the feature importance evaluation rule, the contribution is evaluated and the training data set is reconstructed; Furthermore, based on the feature importance evaluation rule, the contribution is evaluated and the training data set is reconstructed as follows: Arrange the contribution degrees in descending order; The training data of biomarkers corresponding to the top N contributions are selected to reconstruct the training data set.

[0039] Furthermore, the calculation formula for the contribution of each biomarker to the low probability initial prediction model of sepsis is: ; In the formula, Ψ j represents the contribution of the j-th biomarker, S represents the feature subset excluding the j-th biomarker, |S| represents the number of elements in the set S, |N| represents the total number of biomarkers, f(S) represents the prediction result of the low probability initial prediction model for sepsis on the feature subset S, and f(S∪{j}) represents the prediction result of the low probability initial prediction model for sepsis after adding the feature subset of the j-th biomarker to the feature subset S.

[0040] The reconstructed training data set is used to train the initial prediction model for the low probability of sepsis until the corresponding training end conditions are met, thereby obtaining the final prediction model for the low probability of sepsis.

[0041] S4: According to the initial probability of sepsis and the probability of sepsis, the final sepsis prediction result is determined by weight analysis method.

[0042] The final sepsis prediction result is determined by weight analysis, wherein the initial probability of sepsis and the corresponding weights of the sepsis probability are determined, which is not limited in this application, and can be a hierarchical analysis method, an entropy weight method, a principal component analysis method, etc., or mathematical modeling, such as a neural network, etc. The sepsis prediction result and the respective weights of the initial probability of sepsis and the sepsis probability are used as training data to perform model training and obtain a weight assignment model. In specific use, the trained weight assignment model is used to obtain the corresponding weights of the initial probability of sepsis and the sepsis probability.

[0043] According to the initial probability of sepsis and the probability of sepsis, the final sepsis prediction result is determined by weight analysis. The final prediction result obtained in this way not only takes into account the influence of different types of sample data on the prediction result, but also takes into account the more targeted sepsis probability obtained by different prediction models for sample data with high and low probability of sepsis, thereby improving the prediction result of sepsis as a whole.

[0044] Another aspect of the present invention further provides a sepsis risk prediction system based on TIGIT target, which is used to execute any of the above-mentioned sepsis risk prediction methods based on TIGIT target, and the prediction system comprises: An initial sample data acquisition module, used to acquire initial sample data for predicting sepsis; a sepsis initial probability prediction module, connected to the initial sample data acquisition module, and obtaining the initial probability of sepsis by inputting the initial sample data into the first sepsis prediction model; A sepsis probability prediction module is connected to the sepsis initial probability prediction module and is used to determine a method for predicting the sepsis probability based on the comparison between the sepsis initial probability and a preset probability threshold, including the following situations: If the initial probability of sepsis is higher than the preset probability threshold, and TIGIT is greater than the content threshold, it is a high probability of illness, and the high probability of illness sample data is input into the sepsis high probability prediction model to obtain the probability of sepsis; In other cases, the probability of illness is low, and the sample data of low probability of illness are input into the low probability prediction model of sepsis to obtain the probability of sepsis. The output module is connected to the sepsis probability prediction module and is used to determine the final sepsis prediction result through a weight analysis method according to the initial sepsis probability and the sepsis probability.

[0045] It should be noted that the terms used in the present invention are only for describing specific embodiments, rather than limiting the scope of the present application. As shown in the present specification, unless the context clearly indicates an exception, the words "one", "a", "a kind of" and / or "the" do not specifically refer to the singular, but may also include the plural. The terms "include", "comprise" or any other variant thereof are intended to cover non-exclusive inclusion, so that the process, method or device including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, method or device. In the absence of more restrictions, the elements defined by the sentence "include one..." do not exclude the presence of other identical elements in the process, method or device including the elements.

[0046] It should also be noted that the terms "center", "up", "down", "left", "right", "vertical", "horizontal", "inside", "outside", etc., indicating the orientation or positional relationship, are based on the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present invention. Unless otherwise clearly specified and limited, the terms "installed", "connected", "connected", etc. should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be an indirect connection through an intermediate medium, or it can be a connection between the two elements. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0047] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the technical solutions of the embodiments of the present invention.

Claims

1. A method for predicting sepsis risk based on TIGIT targets, characterized in that: The prediction method comprises the following steps: Obtaining initial sample data for predicting sepsis; Inputting the initial sample data into a first sepsis prediction model to obtain an initial probability of sepsis; The method for predicting the probability of sepsis is determined by comparing the initial probability of sepsis with a preset probability threshold, including the following situations: If the initial probability of sepsis is higher than the preset probability threshold, and TIGIT is greater than the content threshold, it is a high probability of illness, and the high probability of illness sample data is input into the sepsis high probability prediction model to obtain the probability of sepsis; In other cases, the probability of illness is low, and the sample data of low probability of illness are input into the low probability prediction model of sepsis to obtain the probability of sepsis. According to the initial probability of sepsis and the probability of sepsis, the final sepsis prediction result was determined by weight analysis method.

2. A method for predicting sepsis risk based on TIGIT target according to claim 1, characterized in that: The initial sample data for predicting sepsis include C-reactive protein, procalcitonin, lactic acid, white blood cell count, platelet count, sequential organ failure score, interleukin 6, interleukin 3, endotoxin, exotoxin, prothrombin time, thromboplastin time, D-dimer, blood creatinine, urine volume, total bilirubin, alanine aminotransferase, and aspartate aminotransferase.

3. A method for predicting sepsis risk based on TIGIT target according to claim 2, characterized in that: The high probability sample data include C-reactive protein, procalcitonin, lactic acid, white blood cell count, platelet count, sequential organ failure score, interleukin 6, interleukin 3, and TIGIT.

4. The method for predicting sepsis risk based on TIGIT target according to claim 2, characterized in that: The low probability of disease sample data include TIGIT, C-reactive protein, endotoxin, exotoxin, prothrombin time, thromboplastin time, D-dimer, blood creatinine, urine volume, total bilirubin, alanine aminotransferase, and aspartate aminotransferase.

5. The method for predicting sepsis risk based on TIGIT target according to claim 1, characterized in that: The method for training the sepsis low probability prediction model is: Acquire historical data of low disease probability samples, wherein the historical data includes multiple groups of low disease probability sample data and the sepsis disease probability corresponding to each group of low disease probability sample data; The low probability sepsis prediction model is trained using the low probability sample historical data until a training end condition is met.

6. The method for predicting sepsis risk based on TIGIT target according to claim 5, characterized in that: Using the low probability sample historical data to train a sepsis low probability prediction model includes the following steps: Obtaining a training data set by standardizing the historical data of the low disease probability samples; Using the training data set to train a low probability sepsis prediction model until a corresponding training end condition is met, thereby obtaining a low probability sepsis initial prediction model; Calculating the contribution of each biomarker in the training data set to the low probability initial prediction model of sepsis according to the characteristic gradient calculation method; Based on the feature importance evaluation rule, the contribution is evaluated and the training data set is reconstructed; The reconstructed training data set is used to train the initial prediction model for the low probability of sepsis until the corresponding training end conditions are met, thereby obtaining the final prediction model for the low probability of sepsis.

7. The method for predicting sepsis risk based on TIGIT target according to claim 6, characterized in that: Based on the feature importance evaluation rule, the contribution is evaluated and the training data set is reconstructed as follows: Arrange the contribution degrees in descending order; The training data of biomarkers corresponding to the top N contributions are selected to reconstruct the training data set.

8. The method for predicting sepsis risk based on TIGIT target according to claim 6, characterized in that: The calculation formula for the contribution of each biomarker to the initial prediction model of low probability of sepsis is: ； In the formula, Ψ j represents the contribution of the j-th biomarker, S represents the feature subset excluding the j-th biomarker, |S| represents the number of elements in the set S, |N| represents the total number of biomarkers, f(S) represents the prediction result of the low probability initial prediction model for sepsis on the feature subset S, and f(S∪{j}) represents the prediction result of the low probability initial prediction model for sepsis after adding the feature subset of the j-th biomarker to the feature subset S.

9. A sepsis risk prediction system based on TIGIT target, used to execute a sepsis risk prediction method based on TIGIT target according to any one of claims 1 to 8, characterized in that: The prediction system comprises: An initial sample data acquisition module, used to acquire initial sample data for predicting sepsis; a sepsis initial probability prediction module, connected to the initial sample data acquisition module, and obtaining the initial probability of sepsis by inputting the initial sample data into the first sepsis prediction model; A sepsis probability prediction module is connected to the sepsis initial probability prediction module and is used to determine a method for predicting the sepsis probability based on the comparison between the sepsis initial probability and a preset probability threshold, including the following situations: If the initial probability of sepsis is higher than the preset probability threshold, and TIGIT is greater than the content threshold, it is a high probability of illness, and the high probability of illness sample data is input into the sepsis high probability prediction model to obtain the probability of sepsis; In other cases, the probability of illness is low, and the sample data of low probability of illness are input into the low probability prediction model of sepsis to obtain the probability of sepsis. The output module is connected to the sepsis probability prediction module and is used to determine the final sepsis prediction result through a weight analysis method according to the initial sepsis probability and the sepsis probability.

Images