Risk prediction method, system and equipment for carbapenem antibiotic treatment and medium
By constructing a kidney injury and death risk prediction model based on characteristic data, combined with machine learning algorithms and correlation rule optimization, the accuracy of carbapenem antibiotic treatment risk prediction is solved, and the provision of personalized treatment plans and dynamic optimization of the model is achieved.
Patent Information
- Application Number
- CN202510461707.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-04-14
- Publication Date
- 2025-08-08
AI Technical Summary
The existing carbapenem antibiotic risk prediction methods have low prediction accuracy in the critically ill patient population and cannot effectively capture multi-factor interactions, resulting in inaccurate early warning information for treatment risk, affecting the safety and effectiveness of clinical applications.
By collecting characteristic data such as the patient's creatinine measurement, urea measurement, drug clearance, continuous renal replacement treatment status, age and albumin measurement, a kidney injury and death risk prediction model was constructed, and key features were screened using univariate logistic regression and SHAP attribution analysis algorithm, combining decision tree and extreme random tree models for training and verification, and building correlation rules to optimize prediction results.
It improves the risk prediction accuracy of carbapenem antibiotic treatment, provides personalized treatment plans, enhances the robustness and prediction accuracy of the model, and dynamically optimizes the model to adapt to changes in real-time clinical data.
Smart Images
Figure CN120452762A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of Internet medical technology, and in particular to a method, system, device and medium for predicting the risk of carbapenem antibiotic treatment. Background Art
[0002] Carbapenem antibiotics occupy a key position in clinical treatment, widely used in the treatment of serious infections due to their potent antibacterial activity. Among them, imipenem and meropenem, as the most commonly used carbapenem antibiotics, play an important role in combating various serious pathogens.
[0003] However, in actual clinical applications, especially for critically ill patients, many urgent problems remain. Critically ill patients' unique pathophysiological conditions, such as organ dysfunction, can severely impact drug metabolism and clearance; significant changes in body weight can alter the drug's volume of distribution; and invasive procedures can trigger a stress response, interfering with drug absorption and transport. These combined factors can lead to significant fluctuations in pharmacokinetic and pharmacodynamic parameters, significantly increasing the risk of treatment failure.
[0004] At the same time, certain carbapenem antibiotics, such as imipenem and meropenem, have the potential for nephrotoxicity, which can easily induce acute kidney injury (AKI) in patients. This risk is particularly pronounced in patients with dynamically changing renal function. Altered renal function can further impair renal excretion of drugs, creating a vicious cycle that seriously threatens patient treatment safety and prognosis.
[0005] Currently, traditional pharmacokinetic / pharmacodynamic models in clinical application primarily focus on predicting treatment efficacy, ignoring the consideration of risk factors during treatment and lacking the ability to effectively predict treatment risks. Existing risk prediction methods for carbapenem antibiotic treatment are mostly based on simple mathematical modeling and analysis based on statistical data. These methods struggle to accurately capture the interactions between multiple factors, resulting in low prediction accuracy and an inability to provide clinicians with reliable and accurate risk warning information. This severely restricts the rational and safe use of carbapenem antibiotics in clinical treatment.
[0006] It can be seen that how to improve the accuracy of risk prediction for carbapenem antibiotic treatment has become a technical problem that needs to be urgently solved by those skilled in the art. Summary of the Invention
[0007] The present invention provides a method and system for predicting the risk of carbapenem antibiotic treatment to solve the technical problem of improving the accuracy of risk prediction for carbapenem antibiotic treatment, thereby improving the accuracy of risk prediction for carbapenem antibiotic treatment and providing doctors with personalized treatment plans based on the results of feature importance analysis.
[0008] In a first aspect, the present invention provides a method for predicting the risk of carbapenem antibiotic treatment, which is applied to predict the risk of imipenem and meropenem treatment, and comprises:
[0009] Collecting characteristic data to be analyzed from patients treated with the carbapenem antibiotic, the characteristic data to be analyzed at least including: creatinine measurement value, urea measurement value / creatinine measurement value, drug clearance rate, continuous renal replacement therapy status, age, and albumin measurement value;
[0010] constructing a first feature data set to be analyzed based on the creatinine measurement value, the urea measurement value / creatinine measurement value, the drug clearance rate, and the continuous renal replacement therapy status; and constructing a second feature data set to be analyzed based on the age, the drug clearance rate, the continuous renal replacement therapy status, and the albumin measurement value;
[0011] Inputting the first feature data set to be analyzed into a pre-built renal injury risk prediction model to obtain a corresponding renal injury risk probability prediction result, wherein the renal injury risk prediction model is trained based on a first data set having the same features as the first feature data set to be analyzed;
[0012] inputting the second feature data set to be analyzed into a pre-built death risk prediction model to obtain a corresponding death risk probability prediction result, wherein the death risk prediction model is trained based on a second data set having the same features as the second feature data set to be analyzed;
[0013] constructing a correlation rule based on an analysis result obtained by performing a correlation analysis on the corresponding damage results in the first data set and the second data set;
[0014] The renal injury risk probability prediction result and the death risk probability prediction result are judged according to the correlation rule, and a risk prediction result is obtained based on the judgment result.
[0015] Preferably, the steps of constructing the renal injury risk prediction model and the death risk prediction model include:
[0016] Obtaining a carbapenem antibiotic treatment risk research dataset, the carbapenem antibiotic treatment risk research dataset including: basic information, laboratory indices, pharmacokinetic / pharmacodynamic parameters, treatment information, and injury outcomes, wherein the basic information at least includes age, the laboratory indices at least include creatinine measurement, urea measurement / creatinine measurement, and albumin measurement, the pharmacokinetic / pharmacodynamic parameters at least include drug clearance, the treatment information at least includes continuous renal replacement therapy status, and the injury outcomes at least include renal injury, death, and no injury;
[0017] screening the pre-processed carbapenem antibiotic treatment risk research data set according to the injury results to obtain a renal injury data group, a death data group, and a non-injury data group, combining the renal injury data group and the non-injury data group into a first data combination, and combining the death data group and the non-injury data group into a second data combination;
[0018] Using a univariate logistic regression algorithm and a SHAP attribution analysis algorithm, the first data combination is screened for key features to obtain a key feature dataset of renal injury risk; using the univariate logistic regression algorithm and the SHAP attribution analysis algorithm, the second data combination is screened for key features to obtain a key feature dataset of death risk;
[0019] constructing the first dataset based on the key feature dataset of renal injury risk and the corresponding injury results, and dividing the first dataset into a first training dataset and a first validation dataset; constructing the second dataset based on the key feature dataset of death risk and the corresponding injury results, and dividing the second dataset into a second training dataset and a second validation dataset;
[0020] Selecting a plurality of machine learning models, wherein the plurality of machine learning models includes at least a decision tree model and an extreme random tree model;
[0021] The first training data set is used to train several of the machine learning models, the first validation data set is used to validate and optimize the trained several machine learning models, and the optimal renal injury risk prediction model is screened out; the second training data set is used to train several of the machine learning models, the second validation data set is used to validate and optimize the trained several machine learning models, and the optimal death risk prediction model is screened out.
[0022] Preferably, the use of a univariate logistic regression algorithm and a SHAP attribution analysis algorithm to screen the first data combination to obtain key features of renal injury risk, and the use of the univariate logistic regression algorithm and the SHAP attribution analysis algorithm to screen the second data combination to obtain key features of death risk include:
[0023] Using each data in the first data combination to train a pre-constructed univariate logistic regression model to obtain a first univariate prediction model corresponding to each data in the first data combination, and using the SHAP attribution analysis method to analyze and screen the first univariate prediction model to obtain key features of renal injury risk;
[0024] Each data in the second data combination is used to train the pre-constructed univariate logistic regression model to obtain a second univariate prediction model corresponding to each data in the second data combination. The second univariate prediction model is analyzed and screened using the SHAP attribution analysis method to obtain key features of death risk.
[0025] Preferably, constructing a correlation rule based on the analysis result obtained by performing a correlation analysis on the corresponding damage results in the first data set and the second data set includes:
[0026] Obtaining a first number of injury results of renal injury in the first dataset, obtaining a second number of injury results of death in the second dataset, and obtaining a total number of samples in the carbapenem antibiotic treatment risk research dataset;
[0027] Obtaining a probability of a renal injury result based on the first number and the total number of samples, and obtaining a probability of a death result based on the second number and the total number of samples;
[0028] A correlation analysis is performed on the renal injury outcome probability and the death outcome probability to construct the correlation criterion.
[0029] Preferably, performing a correlation analysis on the renal injury outcome probability and the death outcome probability to construct the correlation criterion includes:
[0030] A first correlation criterion is obtained based on a size analysis result of the renal injury result probability and the death result probability, wherein the first correlation criterion is that the predicted result of the renal injury risk probability is greater than the predicted result of the death risk probability;
[0031] Calculate a first difference between the probability of the renal injury result and the probability of the death result, set a difference threshold range based on the first difference, and obtain a second correlation criterion based on the difference threshold range, the second correlation criterion being that the second difference between the renal injury risk probability prediction result and the death risk probability prediction result is within the difference threshold range.
[0032] Preferably, judging the renal injury risk probability prediction result and the death risk probability prediction result according to the correlation rule, and obtaining a risk prediction result based on the judgment result, includes:
[0033] Determining whether the renal injury risk probability prediction result and the death risk probability prediction result meet the first correlation criterion; if not, optimizing the renal injury risk prediction model and the death risk prediction model until the renal injury risk probability prediction result and the death risk probability prediction result meet the first correlation criterion;
[0034] Calculate the second difference between the renal injury risk probability prediction result and the death risk probability prediction result, and determine whether the second difference meets the second correlation criterion; if not, optimize the renal injury risk prediction model and the death risk prediction model until the renal injury risk probability prediction result and the death risk probability prediction result meet the second correlation criterion to obtain the risk prediction result.
[0035] Preferably, the feature importance ranking corresponding to the risk prediction result is obtained by the SHAP attribution analysis method;
[0036] The concentration of the carbapenem antibiotic is adjusted according to the risk prediction result and the feature importance ranking.
[0037] In a second aspect, the present invention further provides a carbapenem antibiotic treatment risk prediction system, which implements the above-mentioned carbapenem antibiotic treatment risk prediction method, and the system is applied to the risk prediction of imipenem and meropenem treatment, and the system includes: a data acquisition unit, a feature data set construction unit to be analyzed, a first prediction unit, a second prediction unit, an association rule construction unit, and a prediction result analysis unit;
[0038] The data collection unit is used to collect characteristic data to be analyzed of patients receiving the carbapenem antibiotic treatment, wherein the characteristic data to be analyzed at least includes: creatinine measurement value, urea measurement value / creatinine measurement value, drug clearance rate, continuous renal replacement therapy status, age and albumin measurement value;
[0039] The feature data set to be analyzed constructing unit is configured to construct a first feature data set to be analyzed based on the creatinine measurement value, the urea measurement value / creatinine measurement value, the drug clearance rate, and the continuous renal replacement therapy status, and to construct a second feature data set to be analyzed based on the age, the drug clearance rate, the continuous renal replacement therapy status, and the albumin measurement value;
[0040] The first prediction unit is configured to input the first feature data set to be analyzed into a pre-built renal injury risk prediction model to obtain a corresponding renal injury risk probability prediction result, wherein the renal injury risk prediction model is trained based on a first data set having the same features as the first feature data set to be analyzed;
[0041] The second prediction unit is configured to input the second feature data set to be analyzed into a pre-built death risk prediction model to obtain a corresponding death risk probability prediction result, wherein the death risk prediction model is trained based on a second data set having the same features as the second feature data set to be analyzed;
[0042] The association rule construction unit is configured to construct a correlation rule based on an analysis result obtained by performing a correlation analysis on the corresponding damage results in the first data set and the second data set;
[0043] The prediction result analysis unit is used to judge the renal injury risk probability prediction result and the death risk probability prediction result according to the correlation rule, and obtain a risk prediction result based on the judgment result.
[0044] In a third aspect, the present invention also provides a computer device, comprising a memory, a processor, and a transceiver, which are connected via a bus; the memory is used to store a set of computer program instructions and data, and transmit the stored data to the processor, and the processor executes the program instructions stored in the memory to perform the risk prediction method for carbapenem antibiotic treatment described above.
[0045] In a fourth aspect, the present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed, the risk prediction method for carbapenem antibiotic treatment described above is implemented.
[0046] This application provides a method, system, device, and medium for predicting the risk of carbapenem antibiotic treatment. Compared with the existing technology, the embodiments of this application have the following beneficial effects:
[0047] The risk prediction method for carbapenem antibiotic treatment disclosed in this application integrates pharmacokinetic / pharmacodynamic parameters and clinical indicators to monitor the entire process of carbapenem antibiotic treatment, selects the optimal model for renal injury and death respectively, and dynamically optimizes the prediction results of the two models through the obtained correlation rules, thereby improving the accuracy of risk prediction for carbapenem antibiotic treatment. Based on the results of feature importance analysis, personalized treatment plans are provided to doctors. The model is further dynamically optimized based on real-time collected clinical data to improve the robustness and prediction accuracy of the model. BRIEF DESCRIPTION OF THE DRAWINGS
[0048] Figure 1 This is a schematic diagram of the steps of a method for risk prediction of carbapenem antibiotic treatment provided by a preferred embodiment of the present invention;
[0049] Figure 2 This is a schematic structural diagram of a risk prediction system for carbapenem antibiotic treatment provided by a preferred embodiment of the present invention;
[0050] Figure 3 1 is a diagram showing the internal structure of a computer device according to an embodiment of the present invention. DETAILED DESCRIPTION
[0051] The following is a detailed explanation of the embodiments of the present invention in conjunction with the accompanying drawings. The embodiments are provided for illustrative purposes only and cannot be understood as limitations on the present invention. The accompanying drawings are for reference and illustration purposes only and do not constitute a limitation on the scope of patent protection of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention. In the description of the present invention, the terms "first", "second", "third", etc. are only used for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Therefore, the features defined as "first", "second", "third", etc. may explicitly or implicitly include one or more of the features. In the description of the present invention, unless otherwise specified, the meaning of "multiple" is two or more.
[0052] In the description of the present invention, it should be noted that, unless otherwise expressly specified and limited, the terms "installed", "connected" and "connected" 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 an indirect connection through an intermediate medium, or it can be a communication between the two components. The terms "vertical", "horizontal", "left", "right", "up", "down" and similar expressions used herein are for illustrative purposes only, and do not indicate or imply that the device or component 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. The term "and / or" used herein includes any and all combinations of one or more related listed items. 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.
[0053] In describing the present invention, it should be noted that, unless otherwise defined, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art. The terms used in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. Those skilled in the art will understand the specific meanings of the above terms in the present invention in specific circumstances.
[0054] See also Figure 1 In an embodiment of the present invention, a method for predicting the risk of carbapenem antibiotic treatment is provided, the method comprising:
[0055] S1. Collect characteristic data to be analyzed from patients receiving carbapenem antibiotic treatment, wherein the characteristic data to be analyzed include at least: creatinine measurement value, urea measurement value / creatinine measurement value, drug clearance rate, continuous renal replacement therapy status, age, and albumin measurement value; the characteristic data to be analyzed can be understood as basic risk prediction features obtained by analyzing existing research data sets of patients receiving carbapenem antibiotic treatment, and data that can be used to predict different risks simultaneously based on clinical laboratory data, pharmacokinetic parameter data, pharmacodynamic parameter data, and treatment measures taken by patients receiving carbapenem antibiotic treatment during hospitalization; preferably, in this embodiment, the characteristic data to be analyzed include at least: creatinine measurement value, urea measurement value / creatinine measurement value, drug clearance rate, continuous renal replacement therapy status, age, and albumin measurement value. Among them, continuous renal replacement therapy status includes receiving continuous renal replacement therapy and not receiving continuous renal replacement therapy. In actual application, receiving continuous renal replacement therapy is represented as 1 and not receiving continuous renal replacement therapy is represented as 0. Creatinine measurement values, urea measurement values / creatinine measurement values, drug clearance, age, and albumin measurement values were normalized to ensure that their values were within the range of 0-1, so that all data had the same dimension.
[0056] S2. Based on the creatinine measurement value, the urea measurement value / creatinine measurement value, the drug clearance rate and the continuous renal replacement therapy condition, a first feature data set to be analyzed is constructed, and based on the age, the drug clearance rate, the continuous renal replacement therapy condition and the albumin measurement value, a second feature data set to be analyzed is constructed; in a preferred embodiment of the present application, the risks of carbapenem antibiotic treatment mainly include renal injury and death, wherein the creatinine measurement value, the urea measurement value / creatinine measurement value, the drug clearance rate and the continuous renal replacement therapy condition are used as the feature data set to be analyzed for predicting the risk of renal injury, that is, the first feature data set to be analyzed; and the age, drug clearance rate, continuous renal replacement therapy condition and albumin measurement value are used as the feature data set to be analyzed for predicting the risk of death, that is, the second feature data set to be analyzed. The data factors in the first feature data set to be analyzed and the second feature data set to be analyzed are determined by performing an importance analysis during the model training process.
[0057] S3. Input the first feature data set to be analyzed into a pre-constructed renal injury risk prediction model to obtain a corresponding renal injury risk probability prediction result, wherein the renal injury risk prediction model is trained based on a first data set having the same features as the first feature data set to be analyzed; wherein, in principle, the construction of the renal injury risk prediction model can be obtained by adopting an existing machine learning model construction method that can realize carbapenem antibiotic treatment risk analysis. However, in order to ensure the efficiency, accuracy and generalization of the predictive analysis, in a preferred embodiment of the present application, the carbapenem antibiotic treatment risk research data set is first analyzed to screen the key features of renal injury risk, and then multiple selected machine learning models are trained, verified and screened based on the key features of renal injury risk to obtain the optimal renal injury risk prediction model.
[0058] S4. Input the second feature data set to be analyzed into a pre-constructed death risk prediction model to obtain a corresponding death risk probability prediction result, where the death risk prediction model is trained based on a second data set having the same characteristics as the second feature data set to be analyzed; wherein, in principle, the construction of the death risk prediction model can also be obtained by using an existing machine learning model construction method that can realize carbapenem antibiotic treatment risk analysis. However, in order to ensure the efficiency, accuracy and generalization of the predictive analysis, in a preferred embodiment of the present application, the carbapenem antibiotic treatment risk research data set is first analyzed to screen the key features of the death risk, and then multiple selected machine learning models are trained, verified and screened based on the key features of the death risk to obtain the optimal death risk prediction model.
[0059] Specifically, the steps of constructing the renal injury risk prediction model and the death risk prediction model include:
[0060] A carbapenem antibiotic treatment risk research dataset was obtained, wherein the carbapenem antibiotic treatment risk research dataset includes basic information, laboratory indices, pharmacokinetic / pharmacodynamic parameters, treatment information, and injury results.
[0061] Among them, the basic information includes: age, gender, body mass index and medical history. The medical history mainly includes: hypertension, diabetes, chronic kidney disease, liver disease, skin and soft tissue infection, urinary tract infection, blood infection and pathogen infection, etc., which are diseases that affect kidney damage.
[0062] Laboratory indices included: creatinine measurement, urea measurement, reducing agent measurement, diuretic measurement, albumin measurement, and C-reactive protein measurement;
[0063] Pharmacokinetic / pharmacodynamic parameters include: trough concentration, elimination concentration, clearance, and apparent volume of distribution. Trough concentration is the lowest blood concentration reached after a certain period of drug metabolism and excretion in the body. It is usually the blood concentration measured before the next dose after multiple doses when the drug reaches steady state in the body. Elimination concentration is the blood concentration measured at different time points after the drug enters the elimination phase in the body. Clearance is the apparent volume of distribution of the drug cleared from the body per unit time. It reflects the body's ability to eliminate the drug and is a comprehensive reflection of drug metabolism and excretion, including liver metabolism, renal excretion, and elimination through other pathways. The larger the clearance value, the faster the body clears the drug. The apparent distribution volume value is the volume required for drug distribution when it is assumed that the drug is evenly distributed in the body. It is a theoretical volume concept and does not represent the true physiological volume. The apparent distribution volume can reflect the distribution range of the drug in the body and the degree of binding to the tissue. A large apparent distribution volume value indicates that the drug is widely distributed in the body and may be distributed in large quantities in the tissues. The drug concentration in the plasma is relatively low. A small apparent distribution volume value indicates that the drug is mainly distributed in extracellular fluids such as plasma.
[0064] Treatment information should at least include: medication dosage, medication intervals, and continuous renal replacement therapy status.
[0065] Injury outcomes included renal injury, death, and no injury. Renal injury here specifically referred to acute kidney injury.
[0066] Data preprocessing was performed on the feature data to be analyzed, including missing value processing, deleting variables with a missing rate greater than 90%, and then using modified random forest to fill in the missing values.
[0067] Data standardization processing, Ln standardization is performed on continuous variables to make the data more consistent with the model's assumptions and improve the accuracy of the prediction results.
[0068] Data balance: There is an imbalance between renal injury events and death events. The Borderline SMOTE algorithm (boundary synthetic minority oversampling technology) is used for upsampling to expand the carbapenem antibiotic treatment risk research data set corresponding to death events.
[0069] For the data on disease history and continuous renal replacement therapy in the carbapenem antibiotic treatment risk study dataset, the values are 0 and 1, with 1 indicating presence and 0 indicating absence.
[0070] For numerical data, normalization is performed to ensure that its value is within the range of 0-1, so that all data have the same dimension.
[0071] In this application, a total of 157 patients who received imipenem treatment and 160 patients who received meropenem treatment were selected as patient samples, and the risk research data corresponding to the patient samples were collected to construct a carbapenem antibiotic treatment risk research data set.
[0072] The pre-processed carbapenem antibiotic treatment risk study dataset was screened according to injury results to obtain a renal injury data set, a death data set, and a non-injury data set. The renal injury data set and the non-injury data set were combined into a first data set, and the death data set and the non-injury data set were combined into a second data set. For injury results in specific patient samples, there may be only renal injury, only death, or both renal injury and death. Patient samples with both renal injury and death should be assigned to both the renal injury and death data sets to ensure data adequacy and accuracy.
[0073] Furthermore, a univariate logistic regression algorithm and a SHAP attribution analysis algorithm were used to screen key features of the first data combination to obtain a dataset of key features of renal injury risk. A univariate logistic regression algorithm and a SHAP attribution analysis algorithm were used to screen key features of the second data combination to obtain a dataset of key features of mortality risk. The basis of the logistic regression algorithm is linear regression. The result of linear regression is mapped to a probability value through a logistic function, making it suitable for classification tasks. For univariate logistic regression, that is, with only one independent variable x, its basic formula is:
[0074]
[0075] Where P(Y=1 / x) represents the probability that the dependent variable Y takes the value of 1 given the independent variable x, β0 is the intercept, β1 is the regression coefficient, and e is a natural constant. In a preferred embodiment of the present application, the value of P(Y=1 / x) is P(Y=1 / x)≤0.05.
[0076] Each data in the first data combination is used to train a pre-constructed univariate logistic regression model to obtain a first univariate prediction model corresponding to each data in the first data combination. Each data in the first data combination corresponds to a first univariate prediction model, eliminating the mutual influence between other variables so that the first univariate prediction model can accurately reflect the contribution of the corresponding data to the injury outcome, thereby improving the accuracy of subsequent screening of key features of kidney injury risk. Furthermore, the SHAP attribution analysis algorithm is used to analyze and screen the first univariate prediction model to obtain key features of kidney injury risk. Specifically, the SHAP attribution analysis algorithm is used to analyze the first univariate prediction model corresponding to each data in the first data combination to obtain the SHAP value of each data in the first data combination to measure the importance of each data. The selected key features of kidney injury risk include creatinine measurement value, urea measurement value / creatinine measurement value, drug clearance rate, and continuous renal replacement therapy status.
[0077] For the second data combination, similarly, each data point in the second data combination is used to train the pre-constructed univariate logistic regression model to obtain a second univariate prediction model corresponding to each data point in the second data combination. Each data point in the second data combination corresponds to a second univariate prediction model, eliminating the mutual influence between other variables so that the second univariate prediction model can accurately reflect the contribution of the corresponding data to the death outcome, thereby improving the accuracy of subsequent screening of key features of death risk. Furthermore, the SHAP attribution analysis algorithm is used to analyze and screen the second univariate prediction model to obtain key features of renal injury risk. Specifically, the SHAP attribution analysis algorithm is used to analyze the second univariate prediction model corresponding to each data point in the second data combination to obtain the SHAP value of each data point in the second data combination to measure the importance of each data point. The selected key features of death risk include age, drug clearance, continuous renal replacement therapy status, and albumin measurement value.
[0078] Furthermore, based on the key feature data set of renal injury risk and the corresponding renal injury results, a first data set is constructed, and the first data set is divided into a first training data set and a first validation data set; based on the key feature data set of death risk and the corresponding death results, a second data set is constructed, and the second data set is divided into a second training data set and a second validation data set.
[0079] In a preferred embodiment of the present application, multiple machine learning models are selected, including decision trees, random forests, extreme random trees, XGBoost, support vector machines, naive Bayes, K-neighbor joining algorithm, artificial neural networks, convolutional neural networks, recurrent neural networks, gradient boosting machines and adaptive boosting models.
[0080] The selected machine learning model was trained using the first training data set, and the several trained machine learning models were verified and optimized using the first validation data set. During the validation process, 5-fold cross-validation was used to obtain the AUC, accuracy, sensitivity, specificity, and F1 score results of each machine learning model to judge the performance of each machine learning model and screen out the optimal renal injury risk prediction model. The optimal renal injury risk prediction model screened out in this application was constructed using a decision tree model.
[0081] For the death risk prediction model, similarly, the second training data set is used to train the selected machine learning model, and the second validation data set is used to validate and optimize several trained machine learning models. During the validation process, 5-fold cross-validation is used to obtain the AUC, accuracy, sensitivity, specificity, and F1 score results of each machine learning model, and the performance of each machine learning model is judged to screen out the optimal death risk prediction model. The optimal death risk prediction model screened out in this application is constructed using the extreme random tree model.
[0082] S5. According to the analysis results obtained by performing correlation analysis on the corresponding damage results in the first data set and the second data set, a correlation rule is constructed; in the present application, the number of sample damage results for kidney damage in the first data set is counted to obtain a first number, the number of sample damage results for death in the second data set is counted to obtain a second number, and the number of samples in the initial carbapenem antibiotic treatment risk study data set is counted to obtain the total number of samples. According to the first number and the total number of samples, the probability of kidney damage results is obtained, and according to the second number and the total number of samples, the probability of death results is obtained. The probability of kidney damage results obtained by the above method is the average probability value of the damage result for kidney damage in the carbapenem antibiotic treatment risk study data set, and the probability of death results obtained is the average probability value of the damage result for death in the carbapenem antibiotic treatment risk study data set. Correlation analysis is performed on the probability of kidney damage results and the probability of death results to construct a correlation criterion. First, the probability of kidney damage results and the probability of death results are compared in size. If the probability of kidney damage results is much greater than the probability of death results, then the first correlation criterion is set according to this rule. The first correlation criterion is that the predicted result of the probability of kidney damage risk is greater than the predicted result of the probability of death risk. A large amount of statistical data shows that among patients treated with carbapenem antibiotics, the mortality rate of patients with renal injury is significantly higher than that of patients without renal injury, and there is a positive correlation between the probability of renal injury results and the probability of death results, indicating that the difference between the predicted results of the renal injury risk probability and the predicted results of the death risk probability should be stable within a range. Therefore, the first difference between the renal injury result probability and the death result probability is calculated, and the first difference is used as the middle value and expanded up and down by 5% to obtain the difference threshold range. The second correlation criterion is set according to the difference threshold range. The second correlation criterion is that the second difference between the renal injury risk probability prediction result and the death risk probability prediction result is within the difference threshold range.
[0083] S6. Judge the renal injury risk probability prediction result and the death risk probability prediction result according to the correlation rule, and obtain a risk prediction result based on the judgment result; judge whether the renal injury risk probability prediction result and the death risk probability prediction result meet the first correlation criterion; if not, optimize the renal injury risk prediction model and the death risk prediction model until the renal injury risk probability prediction result and the death risk probability prediction result meet the first correlation criterion; calculate the second difference between the renal injury risk probability prediction result and the death risk probability prediction result, and judge whether the second difference meets the second correlation criterion; if not, optimize the renal injury risk prediction model and the death risk prediction model until the renal injury risk probability prediction result and the death risk probability prediction result meet the second correlation criterion, and output the renal injury risk probability prediction result and the death risk probability prediction result as the risk prediction result.
[0084] In a preferred embodiment of the present application, the SHAP attribution analysis method is further included to obtain the feature importance ranking corresponding to the risk prediction result, and the concentration of the carbapenem antibiotic is adjusted according to the risk prediction result and the feature importance ranking. Specifically, the SHAP attribution analysis method is used to analyze the first feature data set to be analyzed corresponding to the renal injury risk probability prediction result to obtain a first feature importance ranking, and the SHAP attribution analysis method is used to analyze the second feature data set to be analyzed corresponding to the death risk probability prediction result to obtain a second feature importance ranking. Furthermore, based on the first feature importance ranking, the renal injury risk probability prediction result, the second feature importance ranking, and the death risk probability prediction result, the concentration of the carbapenem antibiotic is adjusted, such as by extending the infusion time and reducing the dose, to provide a personalized treatment plan for the patient.
[0085] In this application, during the application of the renal injury risk prediction model and the death risk prediction model, new clinical data are obtained to judge the accuracy of the renal injury risk probability prediction results and the death risk probability prediction results. The renal injury risk prediction model and the death risk prediction model are monitored in real time based on the accuracy judgment results. When the performance declines, the model parameters are automatically adjusted to maintain the prediction accuracy of the renal injury risk prediction model and the death risk prediction model.
[0086] In a preferred embodiment of the present invention, characteristic data to be analyzed of patients receiving carbapenem antibiotic treatment are collected, and the characteristic data to be analyzed include at least: creatinine measurement value, urea measurement value / creatinine measurement value, drug clearance rate, continuous renal replacement therapy status, age and albumin measurement value; a first characteristic data set to be analyzed is constructed based on the creatinine measurement value, urea measurement value / creatinine measurement value, drug clearance rate and continuous renal replacement therapy status, and a second characteristic data set to be analyzed is constructed based on age, drug clearance rate, continuous renal replacement therapy status and albumin measurement value; the first characteristic data set to be analyzed is input into a pre-constructed renal injury risk prediction model to obtain the corresponding renal injury risk prediction model. Risk probability prediction results, the renal injury risk prediction model is trained based on a first data set with the same characteristics as the first feature data set to be analyzed; the second feature data set to be analyzed is input into a pre-constructed death risk prediction model to obtain the corresponding death risk probability prediction results, and the death risk prediction model is trained based on a second data set with the same characteristics as the second feature data set to be analyzed; according to the analysis results obtained by correlation analysis of the corresponding injury results in the first data set and the second data set, a correlation rule is constructed; the renal injury risk probability prediction results and the death risk probability prediction results are judged according to the correlation rules, and the risk prediction results are obtained based on the judgment results. The risk prediction method for carbapenem antibiotic treatment disclosed in this application integrates pharmacokinetic / pharmacodynamic parameters and clinical indicators, monitors the entire process of carbapenem antibiotic treatment, selects the optimal model for renal injury and death, and dynamically optimizes the prediction results of the two models through the obtained correlation rules, thereby improving the risk prediction accuracy of carbapenem antibiotic treatment, providing doctors with personalized treatment plans based on the feature importance analysis results, and further dynamically optimizing the model based on real-time collected clinical data to improve the robustness and prediction accuracy of the model.
[0087] Accordingly, if Figure 2 As shown, based on a risk prediction method for carbapenem antibiotic treatment, an embodiment of the present invention further provides a risk prediction system for carbapenem antibiotic treatment, which implements the risk prediction method for carbapenem antibiotic treatment disclosed in an embodiment of the present invention, including: a data acquisition unit 1, a feature data set to be analyzed construction unit 2, a first prediction unit 3, a second prediction unit 4, an association rule construction unit 5 and a prediction result analysis unit 6;
[0088] The data collection unit 1 is used to collect characteristic data to be analyzed of patients receiving the carbapenem antibiotic treatment, wherein the characteristic data to be analyzed at least includes: creatinine measurement value, urea measurement value / creatinine measurement value, drug clearance rate, continuous renal replacement therapy status, age and albumin measurement value;
[0089] The feature data set construction unit 2 to be analyzed is configured to construct a first feature data set to be analyzed based on the creatinine measurement value, the urea measurement value / creatinine measurement value, the drug clearance rate, and the continuous renal replacement therapy status, and to construct a second feature data set to be analyzed based on the age, the drug clearance rate, the continuous renal replacement therapy status, and the albumin measurement value;
[0090] The first prediction unit 3 is configured to input the first feature data set to be analyzed into a pre-built renal injury risk prediction model to obtain a corresponding renal injury risk probability prediction result, wherein the renal injury risk prediction model is trained based on a first data set having the same features as the first feature data set to be analyzed;
[0091] The second prediction unit 4 is configured to input the second feature data set to be analyzed into a pre-built death risk prediction model to obtain a corresponding death risk probability prediction result, wherein the death risk prediction model is trained based on a second data set having the same features as the second feature data set to be analyzed;
[0092] The association rule construction unit 5 is configured to construct a correlation rule based on the analysis result obtained by performing a correlation analysis on the corresponding damage results in the first data set and the second data set;
[0093] The prediction result analysis unit 6 is used to judge the renal injury risk probability prediction result and the death risk probability prediction result according to the correlation rule, and obtain a risk prediction result based on the judgment result.
[0094] For the specific definition of a risk prediction system for carbapenem antibiotic treatment, please refer to the above-mentioned definition of a risk prediction method for carbapenem antibiotic treatment, which will not be repeated here. Those skilled in the art will appreciate that the various modules and steps described in conjunction with the embodiments disclosed in the present invention can be implemented in hardware, software, or a combination of both. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.
[0095] like Figure 3 As shown, an embodiment of the present invention provides a computer device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, the steps in the risk prediction embodiment of carbapenem antibiotic treatment are implemented, for example Figure 1 Steps S1 to S6 described in .
[0096] Those skilled in the art will understand that the schematic Figure 3 These are merely examples of computer devices and do not constitute limitations on the computer device. The computer device may include more or fewer components than shown in the figure, or a combination of certain components, or different components. For example, the computer device may also include input and output devices, network access devices, buses, etc.
[0097] The processor may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc. The processor is the control center of the computer device, connecting various parts of the entire computer device using various interfaces and lines.
[0098] The memory can be used to store the computer programs and / or modules, and the processor implements various functions of the computer device by running or executing the computer programs and / or modules stored in the memory, and calling the data stored in the memory. The memory can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, at least one application required for a function (such as a sound playback function, an image playback function, etc.); the data storage area can store data created based on the use of the mobile phone (such as audio data, a phone book, etc.). In addition, the memory can include a high-speed random access memory, and can also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), at least one disk storage device, a flash memory device, or other volatile solid-state storage device.
[0099] Wherein, if the module integrated in the computer device is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present invention implements all or part of the process in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and when the computer program is executed by the processor, it can implement the steps of the above-mentioned various method embodiments. Wherein, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may include: any entity or device that can carry the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signal, telecommunication signal and software distribution medium, etc.
[0100] Those skilled in the art will appreciate that all or part of the processes in the above-described method embodiments can be implemented by instructing related hardware through a computer program. The program can be stored in a computer-readable storage medium, and when executed, the program can include the processes in the above-described method embodiments. The storage medium can be a magnetic disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM).
[0101] Accordingly, an embodiment of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium includes a stored computer program, wherein when the computer program is executed, the device where the computer-readable storage medium is located is controlled to perform the steps in the risk prediction of carbapenem antibiotic treatment as described in the above embodiment, for example Figure 1 Steps S1 to S6 described in .
[0102] In summary, the embodiments of the present application provide a method, system, device and medium for predicting the risk of carbapenem antibiotic treatment, which solves the technical problem of improving the accuracy of risk prediction for carbapenem antibiotic treatment. The method comprises: collecting feature data to be analyzed from patients receiving carbapenem antibiotic treatment, the feature data to be analyzed comprising at least: creatinine measurement value, urea measurement value / creatinine measurement value, drug clearance rate, continuous renal replacement therapy status, age and albumin measurement value; constructing a first feature data set to be analyzed based on the creatinine measurement value, urea measurement value / creatinine measurement value, drug clearance rate and continuous renal replacement therapy status, and constructing a second feature data set to be analyzed based on the age, drug clearance rate, continuous renal replacement therapy status and albumin measurement value; and The feature data set to be analyzed is input into a pre-constructed renal injury risk prediction model to obtain a corresponding renal injury risk probability prediction result, and the renal injury risk prediction model is trained based on a first data set having the same characteristics as the first feature data set to be analyzed; the second feature data set to be analyzed is input into a pre-constructed death risk prediction model to obtain a corresponding death risk probability prediction result, and the death risk prediction model is trained based on a second data set having the same characteristics as the second feature data set to be analyzed; according to the analysis results obtained by performing correlation analysis on the corresponding injury results in the first data set and the second data set, a correlation rule is constructed; the renal injury risk probability prediction results and the death risk probability prediction results are judged according to the correlation rule, and a risk prediction result is obtained based on the judgment result. The risk prediction method for carbapenem antibiotic treatment disclosed in this application integrates pharmacokinetic / pharmacodynamic parameters and clinical indicators to monitor the entire process of carbapenem antibiotic treatment, selects the optimal model for renal injury and death respectively, and dynamically optimizes the prediction results of the two models through the obtained correlation rules, thereby improving the accuracy of risk prediction for carbapenem antibiotic treatment. Based on the results of feature importance analysis, personalized treatment plans are provided to doctors. The model is further dynamically optimized based on real-time collected clinical data to improve the robustness and prediction accuracy of the model.
[0103] Each embodiment in this specification is described in a progressive manner, and the same or similar parts of each embodiment can be directly referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment. It should be noted that the various technical features of the above embodiments can be combined arbitrarily. In order to make the description concise, not all possible combinations of the various technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0104] The above-described embodiments merely represent several preferred implementations of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the patent application. It should be noted that a person skilled in the art could make several improvements and substitutions without departing from the technical principles of the present application, and such improvements and substitutions should also be considered within the scope of protection of the present application. Therefore, the scope of protection of the present patent application shall be based on the scope of protection of the claims.
Claims
1. A method for predicting the risk of carbapenem antibiotic treatment, characterized in that: The method is applied to risk prediction of imipenem and meropenem treatment, and the method comprises: Collecting characteristic data to be analyzed from patients treated with the carbapenem antibiotic, the characteristic data to be analyzed at least including: creatinine measurement value, urea measurement value / creatinine measurement value, drug clearance rate, continuous renal replacement therapy status, age, and albumin measurement value; constructing a first feature data set to be analyzed based on the creatinine measurement value, the urea measurement value / creatinine measurement value, the drug clearance rate, and the continuous renal replacement therapy status; and constructing a second feature data set to be analyzed based on the age, the drug clearance rate, the continuous renal replacement therapy status, and the albumin measurement value; Inputting the first feature data set to be analyzed into a pre-built renal injury risk prediction model to obtain a corresponding renal injury risk probability prediction result, wherein the renal injury risk prediction model is trained based on a first data set having the same features as the first feature data set to be analyzed; inputting the second feature data set to be analyzed into a pre-built death risk prediction model to obtain a corresponding death risk probability prediction result, wherein the death risk prediction model is trained based on a second data set having the same features as the second feature data set to be analyzed; constructing a correlation rule based on an analysis result obtained by performing a correlation analysis on the corresponding damage results in the first data set and the second data set; The renal injury risk probability prediction result and the death risk probability prediction result are judged according to the correlation rule, and a risk prediction result is obtained based on the judgment result.
2. The method for predicting the risk of carbapenem antibiotic treatment according to claim 1, wherein: The steps of constructing the renal injury risk prediction model and the death risk prediction model include: Obtaining a carbapenem antibiotic treatment risk research dataset, the carbapenem antibiotic treatment risk research dataset including: basic information, laboratory indices, pharmacokinetic / pharmacodynamic parameters, treatment information, and injury outcomes, wherein the basic information at least includes age, the laboratory indices at least include creatinine measurement, urea measurement / creatinine measurement, and albumin measurement, the pharmacokinetic / pharmacodynamic parameters at least include drug clearance, the treatment information at least includes continuous renal replacement therapy status, and the injury outcomes at least include renal injury, death, and no injury; screening the pre-processed carbapenem antibiotic treatment risk research data set according to the injury results to obtain a renal injury data group, a death data group, and a non-injury data group, combining the renal injury data group and the non-injury data group into a first data combination, and combining the death data group and the non-injury data group into a second data combination; Using a univariate logistic regression algorithm and a SHAP attribution analysis algorithm, the first data combination is screened for key features to obtain a key feature dataset of renal injury risk; using the univariate logistic regression algorithm and the SHAP attribution analysis algorithm, the second data combination is screened for key features to obtain a key feature dataset of death risk; constructing the first dataset based on the key feature dataset of renal injury risk and the corresponding injury results, and dividing the first dataset into a first training dataset and a first validation dataset; constructing the second dataset based on the key feature dataset of death risk and the corresponding injury results, and dividing the second dataset into a second training dataset and a second validation dataset; Selecting a plurality of machine learning models, wherein the plurality of machine learning models includes at least a decision tree model and an extreme random tree model; The first training data set is used to train several of the machine learning models, the first validation data set is used to validate and optimize the trained several machine learning models, and the optimal renal injury risk prediction model is screened out; the second training data set is used to train several of the machine learning models, the second validation data set is used to validate and optimize the trained several machine learning models, and the optimal death risk prediction model is screened out.
3. The method for predicting the risk of carbapenem antibiotic treatment according to claim 2, wherein: The first data combination is screened using a univariate logistic regression algorithm and a SHAP attribution analysis algorithm to obtain key features of renal injury risk. The second data combination is screened using the univariate logistic regression algorithm and the SHAP attribution analysis algorithm to obtain key features of death risk, including: Using each data in the first data combination to train a pre-constructed univariate logistic regression model to obtain a first univariate prediction model corresponding to each data in the first data combination, and using the SHAP attribution analysis method to analyze and screen the first univariate prediction model to obtain key features of renal injury risk; Each data in the second data combination is used to train the pre-constructed univariate logistic regression model to obtain a second univariate prediction model corresponding to each data in the second data combination. The second univariate prediction model is analyzed and screened using the SHAP attribution analysis method to obtain key features of death risk.
4. The method for predicting the risk of carbapenem antibiotic treatment according to claim 2, wherein: The constructing of correlation rules based on the analysis results obtained by performing correlation analysis on the corresponding damage results in the first data set and the second data set includes: Obtaining a first number of injury results of renal injury in the first dataset, obtaining a second number of injury results of death in the second dataset, and obtaining a total number of samples in the carbapenem antibiotic treatment risk research dataset; Obtaining a probability of a renal injury result based on the first number and the total number of samples, and obtaining a probability of a death result based on the second number and the total number of samples; A correlation analysis is performed on the renal injury outcome probability and the death outcome probability to construct the correlation criterion.
5. The method for predicting the risk of carbapenem antibiotic treatment according to claim 4, wherein: The performing of correlation analysis on the renal injury result probability and the death result probability to construct the correlation criterion includes: A first correlation criterion is obtained based on a size analysis result of the renal injury result probability and the death result probability, wherein the first correlation criterion is that the predicted result of the renal injury risk probability is greater than the predicted result of the death risk probability; Calculate a first difference between the probability of the renal injury result and the probability of the death result, set a difference threshold range based on the first difference, and obtain a second correlation criterion based on the difference threshold range, the second correlation criterion being that the second difference between the renal injury risk probability prediction result and the death risk probability prediction result is within the difference threshold range.
6. The method for predicting the risk of carbapenem antibiotic treatment according to claim 5, wherein: The step of judging the renal injury risk probability prediction result and the death risk probability prediction result according to the correlation rule, and obtaining a risk prediction result based on the judgment result, includes: Determining whether the renal injury risk probability prediction result and the death risk probability prediction result meet the first correlation criterion; if not, optimizing the renal injury risk prediction model and the death risk prediction model until the renal injury risk probability prediction result and the death risk probability prediction result meet the first correlation criterion; Calculate the second difference between the renal injury risk probability prediction result and the death risk probability prediction result, and determine whether the second difference meets the second correlation criterion; if not, optimize the renal injury risk prediction model and the death risk prediction model until the renal injury risk probability prediction result and the death risk probability prediction result meet the second correlation criterion to obtain the risk prediction result.
7. The method for predicting the risk of carbapenem antibiotic treatment according to claim 3, wherein: The method further comprises: Obtaining the feature importance ranking corresponding to the risk prediction result through the SHAP attribution analysis method; The concentration of the carbapenem antibiotic is adjusted according to the risk prediction result and the feature importance ranking.
8. A carbapenem antibiotic treatment risk prediction system, which implements the carbapenem antibiotic treatment risk prediction method according to any one of claims 1 to 7, characterized in that: The system is applied to the risk prediction of imipenem and meropenem treatment, and the system comprises: a data acquisition unit, a feature data set construction unit to be analyzed, a first prediction unit, a second prediction unit, an association rule construction unit, and a prediction result analysis unit; The data collection unit is used to collect characteristic data to be analyzed of patients receiving the carbapenem antibiotic treatment, wherein the characteristic data to be analyzed at least includes: creatinine measurement value, urea measurement value / creatinine measurement value, drug clearance rate, continuous renal replacement therapy status, age and albumin measurement value; The feature data set to be analyzed constructing unit is configured to construct a first feature data set to be analyzed based on the creatinine measurement value, the urea measurement value / creatinine measurement value, the drug clearance rate, and the continuous renal replacement therapy status, and to construct a second feature data set to be analyzed based on the age, the drug clearance rate, the continuous renal replacement therapy status, and the albumin measurement value; The first prediction unit is configured to input the first feature data set to be analyzed into a pre-built renal injury risk prediction model to obtain a corresponding renal injury risk probability prediction result, wherein the renal injury risk prediction model is trained based on a first data set having the same features as the first feature data set to be analyzed; The second prediction unit is configured to input the second feature data set to be analyzed into a pre-built death risk prediction model to obtain a corresponding death risk probability prediction result, wherein the death risk prediction model is trained based on a second data set having the same features as the second feature data set to be analyzed; The association rule construction unit is configured to construct a correlation rule based on an analysis result obtained by performing a correlation analysis on the corresponding damage results in the first data set and the second data set; The prediction result analysis unit is used to judge the renal injury risk probability prediction result and the death risk probability prediction result according to the correlation rule, and obtain a risk prediction result based on the judgment result.
9. A computer device, characterized in that: The computer device includes a memory, a processor, and a transceiver, which are connected via a bus; the memory is used to store a set of computer program instructions and data, and transmit the stored data to the processor, and the processor executes the program instructions stored in the memory to perform the risk prediction method for carbapenem antibiotic treatment according to any one of claims 1 to 7.
10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed, the method for predicting the risk of carbapenem antibiotic treatment according to any one of claims 1 to 7 is implemented.