Personalized prediction system for clinical curative effect of phlegm-heat clearing injection

Abstract

The invention provides a personalized prediction system for the clinical curative effect of a phlegm-heat clearing injection. The personalized prediction system comprises an input module, a curative effect prediction module and an output module. The input module is used for acquiring the basic condition of a to-be-predicted patient, the combined medication information and the injection time of the phlegm-heat clearing injection; the curative effect prediction module is used for predicting the ideal probability value of the clinical curative effect of a to-be-predicted patient under different combined medication treatment schemes by using the established personalized prediction model for the clinical curative effect of the phlegm-heat clearing injection by taking the C-reactive protein as an outcome index; and the output module is used for outputting probability values of ideal clinical curative effects of the to-be-predicted patient under different combined medication treatment schemes. Based on the screened 13 indexes, the XGBoost model is adopted to construct the personalized prediction model and the prediction system for the clinical curative effect of the phlegm-heat clearing injection, and the optimal personalized treatment scheme of the phlegm-heat clearing injection and the combined drug thereof is rapidly, accurately and efficiently determined.

Description

Technical Field

[0001] The present invention relates to the technical field of machine learning, and in particular to a personalized prediction system for the clinical efficacy of Tanreqing injection. Background Art

[0002] Tanreqing injection is composed of five medicinal materials: bear bile powder, goat horn, scutellaria, forsythia, and honeysuckle. It has the effects of clearing heat, resolving phlegm, and detoxifying. It is used for wind-heat lung heat disease with phlegm-heat blocking the lung syndrome, with symptoms of fever, cough, uncomfortable expectoration, sore throat, thirst, red tongue, and yellow fur; early pneumonia, acute bronchitis, acute exacerbation of chronic bronchitis, and upper respiratory tract infection. At present, the relevant research on the effectiveness of Tanreqing injection includes RCT research, retrospective clinical research, etc., and lacks personalized prediction. The present invention collects multi-center real-world data, combines artificial intelligence algorithms, and establishes a prediction model for the efficacy of Tanreqing injection in the treatment of patients with lung infections, providing a reference for the formulation of clinical medication plans. Summary of the invention

[0003] In order to overcome the above technical defects, the object of the present invention is to provide a personalized prediction system for the clinical efficacy of Tanreqing injection, which comprises:

[0004] An input module, wherein the input module is used to obtain the basic information of the patient to be predicted, the combined medication information, and the injection time of Tanreqing injection; wherein the basic information includes age, gender, albumin baseline value, albumin / globulin baseline value, mean corpuscular hemoglobin concentration baseline value, red blood cell distribution width coefficient of variation baseline value, hemoglobin amount, and uric acid baseline value; wherein the combined medication information is a combined medication treatment plan consisting of β-lactams, quinolones, β-lactamase inhibitor compound preparations, and carbapenem antibiotics;

[0005] The efficacy prediction module is used to use C-reactive protein as the outcome indicator and the constructed personalized clinical efficacy prediction model of Tanreqing injection to predict the probability value of the ideal clinical efficacy of the predicted patient under different combined drug treatment schemes;

[0006] The output module is used to output the probability value of the ideal clinical efficacy of the patient to be predicted under different combined drug treatment schemes. The combined drug treatment scheme corresponding to the highest probability value is the optimal combined drug treatment scheme for the patient to be predicted.

[0007] Furthermore, the method for constructing the personalized prediction model for clinical efficacy of Tanreqing injection comprises the following steps:

[0008] Step S1: Collect data of several inpatients who used Tanreqing injection and had C-reactive protein information from HIS databases of multiple hospitals and divide them into training set and validation set;

[0009] Step S2: Feature engineering: (1) Determine the outcome indicator: C-reactive protein is selected as the evaluation indicator of treatment outcome. If the last test result before discharge is high and the highest value during hospitalization, it is defined as disease progression, and other patients are defined as disease remission; (2) Analyze the factors affecting the efficacy of Tanreqing injection from three perspectives: the patient himself, the usage, dosage and medication time of Tanreqing injection, and combined medication. The combined medication includes β-lactamase inhibitors, quinolones, β-lactamase inhibitor combination preparations, and carbapenem antibiotics;

[0010] Step S3: Screening prediction models: Use GBDT, XGBoost, LIghtGBM, Adaboost, and Catboost in the machine learning models to perform preliminary calculations on the cleaned data, use exhaustive search GridSearch to adjust parameters, and then use the five indicators of Precision, Recall, F1-score, AUC, and Accuracy to evaluate the model effects of GBDT, XGBoost, LIghtGBM, Adaboost, and Catboost, so as to screen out the XGBoost model with the best effect;

[0011] Step S4: Use the feature sequence forward selection algorithm based on the random forest algorithm to perform feature screening. The algorithm starts searching from an empty set and adds one feature to the feature subset each time, so that the area under the ROC curve of the evaluation function AUC reaches the optimal value. When the number of introduced variables reaches 8, the model can reach the local optimum.

[0012] Step S5: Using the training set to train the XGBoost model, eight indicators with higher SHAP values ​​were selected, including white-globulin ratio, duration of use of Tanreqing injection, albumin, mean corpuscular hemoglobin concentration, coefficient of variation of red blood cell distribution width, hemoglobin amount, uric acid, and patient age;

[0013] Step S6: On the basis of the 8 indicators, β-lactams, quinolones, β-lactamase inhibitor combination preparations and carbapenems as well as patient gender were added, and the XGBoost model was retrained based on these 13 indicators to finally construct a personalized prediction model for the clinical efficacy of Tanreqing Injection.

[0014] Further, in the input module, the albumin baseline value, the albumin / globulin baseline value, the mean corpuscular hemoglobin concentration baseline value, the red blood cell distribution width coefficient of variation baseline value, the hemoglobin amount and the uric acid baseline value are divided into normal, high and / or low based on clinical reference values.

[0015] Furthermore, in the input module, β-lactams, quinolones, β-lactamase inhibitor combination preparations, and carbapenem antibiotics are divided into two situations: used and not used.

[0016] Furthermore, in the input module, age and Tanreqing injection injection time are both continuous variables, and the unit of Tanreqing injection injection time is day.

[0017] Compared with the prior art, the above technical solution has the following beneficial effects:

[0018] This application is based on the 13 screened indicators and uses the XGBoost model to construct a personalized prediction model for the clinical efficacy of Tanreqing injection and its prediction system, which can predict the clinical efficacy of Tanreqing injection for specific patients under different combination drug treatment regimens, thereby achieving rapid, accurate and efficient determination of the optimal personalized treatment plan for Tanreqing injection and its combination drug. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 This is a module structure diagram of the personalized prediction system for clinical efficacy of Tanreqing injection of this application;

[0020] Figure 2 This is a graph of feature screening results using the Sequential Forward Selection (SFS) algorithm based on the random forest algorithm;

[0021] Figure 3 This is the SHAP graph obtained after preliminary training using the XGBoost model;

[0022] Figure 4 This is an example diagram of the user interface of the personalized prediction system for the clinical efficacy of Tanreqing Injection in this application. DETAILED DESCRIPTION

[0023] The advantages of the present invention are further described below in conjunction with the accompanying drawings and specific embodiments. Those skilled in the art should understand that the following specific description is illustrative rather than restrictive, and should not be used to limit the scope of protection of the present invention.

[0024] like Figure 1 As shown, this embodiment provides a personalized prediction system for the clinical efficacy of Tanreqing injection, which includes: an input module, an efficacy prediction module and an output module.

[0025] The input module is used to obtain the basic information of the patient to be predicted, the combined medication information, and the injection time of Tanreqing injection. Among them, the basic information includes age, gender, albumin baseline value, albumin / globulin baseline value, mean corpuscular hemoglobin concentration baseline value, red blood cell distribution width coefficient of variation baseline value, hemoglobin amount and uric acid baseline value. Among them, the combined medication information is a combined medication treatment plan consisting of β-lactams, quinolones, β-lactamase inhibitor compound preparations, and carbapenem antibiotics. The albumin baseline value, albumin / globulin baseline value, mean corpuscular hemoglobin concentration baseline value, red blood cell distribution width coefficient of variation baseline value, hemoglobin amount and uric acid baseline value are divided into normal, high and / or low based on the clinical reference value. β-lactams, quinolones, β-lactamase inhibitor compound preparations, and carbapenem antibiotics are divided into two situations: use and non-use. Age and Tanreqing injection injection time are both continuous variables, and the unit of Tanreqing injection injection time is day.

[0026] Taking C-reactive protein as the outcome indicator, the constructed personalized prediction model for the clinical efficacy of Tanreqing Injection was used to predict the probability value of ideal clinical efficacy of the predicted patients under different combination drug treatment regimens.

[0027] The method for constructing a personalized prediction model for the clinical efficacy of Tanreqing injection includes the following steps:

[0028] Step S1: Data of several hospitalized patients who used Tanreqing injection and had C-reactive protein information were collected from the HIS databases of multiple hospitals and divided into a training set and a validation set.

[0029] Specifically, the case information of 83,785 inpatients from the HIS database of 10 tertiary hospitals across the country from 2015 to 2023 was collected, and the data of 20,489 patients using Tanreqing injection were screened out, and the data of 2,378 patients with missing CRP outcome indicators were excluded. The remaining 18,111 inpatients who used Tanreqing injection and had C-reactive protein information were used to construct a data set, which was divided into a training set (14,489 cases) and a validation set (3,622 cases). The data information of each patient included but was not limited to demographic indicators (including but not limited to age and gender, etc.), patient laboratory test baseline values ​​(including but not limited to albumin baseline value, albumin / globulin baseline value, mean corpuscular hemoglobin concentration baseline value, red blood cell distribution width coefficient of variation baseline value, hemoglobin amount and uric acid baseline value, etc.), combined use of antibacterial drugs (including but not limited to β-lactams, quinolones, β-lactamase inhibitor combination preparations, carbapenem antibiotics, etc.) and Tanreqing injection medication time (unit: "day"), etc.

[0030] Step S2: Feature Engineering:

[0031] (1) Determine the outcome indicator: C-reactive protein was selected as the evaluation indicator of treatment outcome. If the last test result before discharge was high and was the highest value during hospitalization, it was defined as disease progression (0); other patients were defined as disease remission (1).

[0032] (2) Analyze the factors that affect the efficacy of Tanreqing Injection from three perspectives: the patient himself, the usage, dosage and time of Tanreqing Injection, and combined medication.

[0033] Baseline status of patients: Demographic characteristics of patients include age and gender. The first laboratory test results of patients after admission are used as indicators to evaluate the baseline status of patients, and are divided into three groups: low, normal, and high according to the normal value range specified in clinical guidelines.

[0034] Since most patients in this study used 20-40 ml (91.8%) of Tanreqing injection at a time, and 99.7% of the cases used the injection qd, we only included the duration of Tanreqing injection in the analysis. In terms of combined medication, we selected the four most frequently used antimicrobial drugs in this study for analysis: β-lactamase inhibitors, quinolones, β-lactamase inhibitor compound preparations, and carbapenems. The actual use of combined medication includes several situations where one or more of β-lactamase inhibitors, quinolones, β-lactamase inhibitor compound preparations, and carbapenems are not used, or none of the four drugs are used.

[0035] Step S3: Screening prediction models: Use GBDT, XGBoost, LIghtGBM, Adaboost, and Catboost in the machine learning models to perform preliminary calculations on the cleaned data. We chose the exhaustive search (GridSearch) in the parameter adjustment method: select the best performing parameters through loop traversal. Then, the model effects of GBDT, XGBoost, LIghtGBM, Adaboost, and Catboost were evaluated with the five indicators of Precision, Recall, F1-score, AUC, and Accuracy, as shown in Table 1, so as to screen out the XGBoost model with the best effect.

[0036] Table 1 Model performance comparison

[0037] Model Precision Recall F1-score AUC Accuracy GBDT 0.6961 0.5621 0.6224 0.6896 0.7079 XGBoost 0.6946 0.6243 0.6576 0.7096 0.7219 LightGBM 0 0 0 0.5 0.5722 Adaboost 0.6785 0.5347 0.5981 0.6727 0.6926 Catboost 0.6149 0.5781 0.5959 0.6537 0.6646

[0038] By comparing the five indicators in Table 1, it can be seen that the XGBoost model has the best performance. Therefore, the XGBoost model will be used for training and to build a personalized prediction model for the clinical efficacy of Tanreqing Injection.

[0039] Step S4: Use the Sequential Forward Selection (SFS) algorithm based on the random forest algorithm to screen features. The algorithm starts searching from an empty set and adds a feature to the feature subset each time to make the evaluation function (AUC: area under the ROC curve) reach the optimal value. If the candidate feature subset is not as good as the feature subset of the previous round, then stop the iteration and use the feature subset of the previous round as the optimal feature selection result. The results are as follows: Figure 2 We found that the model can reach local optimality when the number of introduced variables reaches 8. Because the XGBoost model can process high-dimensional sparse features in a distributed manner, there is no need to consider the collinearity problem between variables.

[0040] Step S5: Use the training set to train the XGBoost model, and select eight indicators with higher SHAP values, including white-globulin ratio, Tanreqing injection medication time, albumin, mean corpuscular hemoglobin concentration, red blood cell distribution width coefficient of variation, hemoglobin amount, uric acid, and patient age, according to the SHAP value, which have higher contribution to the model. Figure 3 :Each point in the figure is a case. The horizontal axis represents the SHAP value, which represents the contribution of the variable to the model (the degree of influence on the outcome indicator), and the SHAP value is greater than 0 (the point is on the right), indicating that the variable contributes positively to the model, otherwise it is negative. The vertical axis is the true value of the variable. The red color indicates that the higher the value of the variable, and the bluer the color, the lower the true value of the variable. ALB / GLO (albumin-globulin ratio), RDW_CV (coefficient of variation of red blood cell volume), HB (hemoglobin): The red points are concentrated on the left, and the blue points are concentrated on the right, indicating that patients with low initial values ​​of the above indicators are more likely to have disease remission. ALB (albumin), MCHC (mean corpuscular hemoglobin concentration), UA (uric acid): The blue points are concentrated on the left, indicating that patients with low values ​​of the above indicators are more likely to have disease progression. Time (time of taking Tanreqing injection), Age (patient age): The blue points are concentrated on the left, indicating that patients with shorter medication time and younger age are more likely to have disease progression.

[0041] Step S6: On the basis of the 8 indicators, β-lactams, quinolones, β-lactamase inhibitor combination preparations and carbapenems as well as patient gender are added, and the XGBoost model is retrained based on these 13 indicators to finally build a personalized prediction model for the clinical efficacy of Tanreqing injection, and a personalized prediction system for the clinical efficacy of Tanreqing injection is developed based on the personalized prediction model for the clinical efficacy of Tanreqing injection, and the model is visualized. For example, the user interface of the personalized prediction system for the clinical efficacy of Tanreqing injection is as follows Figure 4 shown.

[0042] Output the probability value of the ideal clinical efficacy of the patient to be predicted under different combination drug treatment schemes. The combination drug treatment scheme corresponding to the highest probability value is the optimal combination drug treatment scheme for the patient to be predicted.

[0043] based on Figure 4 In the prediction platform, when inputting information, age and medication time (unit: "day") are continuous variables; gender is divided into "male" and "female"; albumin baseline value, albumin / globulin baseline value, mean corpuscular hemoglobin concentration baseline value, red blood cell distribution width coefficient of variation baseline value, hemoglobin amount and uric acid baseline value are divided into "normal", "high" and / or "low" based on clinical reference values; each drug among β-lactams, quinolones, β-lactamase inhibitor combination preparations and carbapenem antibiotics is divided into "used" and "not used".

[0044] For a particular patient, Figure 4 The basic information (age and gender) and laboratory test results (baseline albumin value, albumin / globulin value, mean corpuscular hemoglobin concentration baseline value, red blood cell distribution width coefficient of variation baseline value, hemoglobin amount and uric acid baseline value) are fixed. By constantly changing the specific circumstances of combined medication (β-lactams, quinolones, β-lactamase inhibitor combination preparations, carbapenem antibiotics) and the use time of Tanreqing injection, the ideal prediction value of efficacy under different treatment plans is output. Among them, the combined medication treatment plan corresponding to the highest probability value is the optimal combined medication treatment plan for the patient to be predicted, thereby realizing personalized prediction of the clinical efficacy of Tanreqing injection.

[0045] It should be noted that the embodiments of the present invention have better practicability and do not impose any form of limitation on the present invention. Any technician familiar with the field may use the technical content disclosed above to change or modify it into an equivalent effective embodiment. However, any modification or equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention are still within the scope of the technical solution of the present invention.

Claims

1. A personalized prediction system for clinical efficacy of Tanreqing injection, characterized in that: include: An input module, wherein the input module is used to obtain the basic information of the patient to be predicted, the combined medication information, and the injection time of Tanreqing injection; wherein the basic information includes age, gender, albumin baseline value, albumin / globulin baseline value, mean corpuscular hemoglobin concentration baseline value, red blood cell distribution width coefficient of variation baseline value, hemoglobin amount, and uric acid baseline value; wherein the combined medication information is a combined medication treatment plan consisting of β-lactams, quinolones, β-lactamase inhibitor compound preparations, and carbapenem antibiotics; The efficacy prediction module is used to use C-reactive protein as the outcome indicator and the constructed personalized clinical efficacy prediction model of Tanreqing injection to predict the probability value of the ideal clinical efficacy of the predicted patient under different combined drug treatment schemes; The output module is used to output the probability value of the ideal clinical efficacy of the patient to be predicted under different combined drug treatment schemes.

2. The personalized prediction system for clinical efficacy of Tanreqing injection according to claim 1, characterized in that: The method for constructing a personalized prediction model for the clinical efficacy of Tanreqing injection comprises the following steps: Step S1: Collect data of several inpatients who used Tanreqing injection and had C-reactive protein information from HIS databases of multiple hospitals and divide them into training set and validation set; Step S2: Feature engineering: (1) Determine the outcome indicator: C-reactive protein is selected as the evaluation indicator of treatment outcome. If the last test result before discharge is high and the highest value during hospitalization, it is defined as disease progression, and other patients are defined as disease remission; (2) Analyze the factors affecting the efficacy of Tanreqing injection from three perspectives: the patient himself, the usage, dosage and medication time of Tanreqing injection, and combined medication. The combined medication includes β-lactamase inhibitors, quinolones, β-lactamase inhibitor combination preparations, and carbapenem antibiotics; Step S3: Screening prediction models: Use GBDT, XGBoost, LIghtGBM, Adaboost, and Catboost in the machine learning models to perform preliminary calculations on the cleaned data, use exhaustive search GridSearch to adjust parameters, and then use the five indicators of Precision, Recall, F1-score, AUC, and Accuracy to evaluate the model effects of GBDT, XGBoost, LIghtGBM, Adaboost, and Catboost, so as to screen out the XGBoost model with the best effect; Step S4: Use the feature sequence forward selection algorithm based on the random forest algorithm to perform feature screening. The algorithm starts searching from an empty set and adds one feature to the feature subset each time, so that the area under the ROC curve of the evaluation function AUC reaches the optimal value. When the number of introduced variables reaches 8, the model can reach the local optimum. Step S5: Using the training set to train the XGBoost model, eight indicators with higher SHAP values ​​were selected, including white-globulin ratio, duration of use of Tanreqing injection, albumin, mean corpuscular hemoglobin concentration, coefficient of variation of red blood cell distribution width, hemoglobin amount, uric acid, and patient age; Step S6: On the basis of the 8 indicators, β-lactams, quinolones, β-lactamase inhibitor combination preparations and carbapenems as well as patient gender were added, and the XGBoost model was retrained based on these 13 indicators to finally construct a personalized prediction model for the clinical efficacy of Tanreqing Injection.

3. The personalized prediction system for clinical efficacy of Tanreqing injection according to claim 1, characterized in that: In the input module, the baseline values ​​of albumin, albumin / globulin, mean corpuscular hemoglobin concentration, coefficient of variation of red cell distribution width, hemoglobin amount, and uric acid were classified as normal, high, and / or low based on clinical reference values.

4. The personalized prediction system for clinical efficacy of Tanreqing injection according to any one of claims 1 to 3, characterized in that: In the input module, β-lactams, quinolones, β-lactamase inhibitor combination preparations, and carbapenem antibiotics were divided into two situations: used and not used.

5. The personalized prediction system for clinical efficacy of Tanreqing injection according to any one of claims 1 to 3, characterized in that: In the input module, age and time of injection of Tanreqing injection are both continuous variables, and the unit of time of injection of Tanreqing injection is day.

Images