CN116072259B - An optimal dose selection method and system for neonatal β-lactam drugs - Google Patents

Abstract

The invention discloses a method and system for selecting the optimal dosage of β-lactam drugs for newborns, which includes: obtaining the characteristic information of the newborn, selecting drugs, and drug dosage and frequency information; selecting drugs according to the characteristic information of the newborn, The drug dosage and frequency information and the trained drug optimal dose selection model determine whether the drug dosage is the recommended optimal dose. Among them, the drug optimal dose selection model uses the newborn's characteristic information, selected drugs, The drug dosage and frequency information is the input, and whether the drug dosage is the recommended optimal dose is the output, which is obtained through CatBoost model construction; when the drug dosage is not the recommended optimal dose, the drug dosage is readjusted. Dosage, so that the adjusted drug dosage is determined to be the recommended optimal dose through the drug optimal dosage selection model. Accurate prediction of optimal drug dosage is achieved.

Description

An optimal dose selection method and system for neonatal β-lactam drugs

Technical field

The present invention relates to the technical field of optimal dose prediction of drugs, and in particular to a method and system for optimal dose selection of neonatal β-lactam drugs.

Background technique

The statements in this section merely provide background technical information related to the present invention and do not necessarily constitute prior art.

Neonatal sepsis is a systemic disease with clinical manifestations ranging from subclinical infection to severe local or systemic infection, with high morbidity and mortality. β-lactams are the most commonly used antibiotics to treat neonatal sepsis. one. How to accurately predict the optimal dose of β-lactam drugs is currently a problem that needs to be solved.

The optimal dose of beta-lactams is determined by multiple factors, including: (1) Rapid physiological changes and unique pathophysiology of neonates lead to wide inter- and intra-individual variability in drug disposition and clinical response sex; (2) Infectious entities and their corresponding minimum inhibitory concentrations (MICs) are an important basis for antibiotic treatment. However, it is difficult to obtain positive culture results and antimicrobial sensitivity results of neonatal bacteria in clinical practice; (3) The pharmacodynamic (PD) target for time-dependent β-lactams is the portion of time during which the free antibiotic concentration remains above the minimum inhibitory concentration of the target pathogen (%fT > MIC). However, specific PD target values for neonates are Controversially, there is no universally accepted PD target, as in recent neonatal studies, PD targets ranged from 40% fT>MIC to 100% fT>4-5xMIC; (4) Central differences, specific β-lactams The choice of drugs and dosing regimen chosen for neonatal sepsis vary between centers and regions.

At present, population pharmacokinetic models are mainly used to predict the optimal dose of β-lactam drugs. However, when predicting the optimal dose of β-lactam drugs, the PD index of the drug is not considered, making the prediction results Inaccurate.

Contents of the invention

In order to solve the above problems, the present invention proposes a method and system for selecting the optimal dose of β-lactam drugs for newborns, which can accurately predict the optimal dose of β-lactam drugs.

In order to achieve the above objects, the present invention adopts the following technical solutions:

In the first aspect, a method for optimal dosage selection of neonatal β-lactam drugs is proposed, including:

Obtain the newborn's characteristic information, selected drugs, and drug dosage and frequency information;

Based on the newborn's characteristic information, selected drugs, drug dosage and frequency information, and the trained drug optimal dose selection model, determine whether the drug dosage is the recommended optimal dose. Among them, the drug optimal dose selection model, Taking the newborn's characteristic information, selected drugs, drug dosage and frequency information as input, and taking whether the drug dosage is the recommended optimal dose as the output, it is obtained through CatBoost model construction;

When the drug dosage is not the recommended optimal dose, the drug dosage is readjusted so that the adjusted drug dosage is determined to be the recommended optimal dose through the optimal drug dosage selection model.

In the second aspect, an optimal dosage selection system for neonatal β-lactam drugs is proposed, including:

The data acquisition module is used to obtain the characteristic information of newborns, selected drugs, and drug dosage and frequency information;

The recommended optimal dose determination module is used to determine whether the drug dosage is the optimal dose based on the newborn's characteristic information, selected drugs, drug dosage and frequency information, and the trained drug optimal dose selection model, where , the optimal drug dose selection model takes the newborn's characteristic information, selected drugs, drug dosage and frequency information as input, and uses whether the drug dosage is the optimal dose as the output, which is obtained through the CatBoost model construction; when the drug is given When the drug dosage is not the recommended optimal dose, the drug dosage is readjusted so that the adjusted drug dosage is determined to be the recommended optimal dose through the optimal drug dosage selection model.

In a third aspect, an electronic device is proposed, including a memory and a processor, and computer instructions stored in the memory and run on the processor. When the computer instructions are run by the processor, a neonatal beta-lactam drug is completed. The steps described in the optimal dose selection method.

In a fourth aspect, a computer-readable storage medium is proposed, which is used to store computer instructions. When the computer instructions are executed by a processor, the steps described in a method for selecting an optimal dose of beta-lactam drugs for newborns are completed.

Compared with the prior art, the beneficial effects of the present invention are:

1. When predicting the optimal dose of a drug, the present invention comprehensively considers the characteristic information of the newborn and the selected drug, and determines the relationship between the PD index and the PD target of the drug under the dosage. Whether the dosage is the recommended best dose, and then select the minimum value from the recommended best dose as the best dose of the drug, achieving accurate prediction of the best dose of the drug.

2. When selecting the PD target, the present invention takes the MICs of all pathogenic bacteria of sepsis into consideration, so that the selected optimal dose can be effective against all pathogenic bacteria of sepsis.

3. The present invention only realizes the prediction of the optimal dose of recommended drugs through the drug optimal dose selection model, and improves the prediction efficiency on the basis of ensuring the accuracy of optimal dose prediction.

Additional advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention.

Description of the drawings

The description and drawings that constitute a part of this application are used to provide a further understanding of this application. The illustrative embodiments and their descriptions of this application are used to explain this application and do not constitute an improper limitation of this application.

Figure 1 is a flow chart of the method disclosed in Embodiment 1;

Figure 2 is an application flow chart of the optimal drug dosage selection model disclosed in Example 1.

Detailed ways

The present invention will be further described below in conjunction with the accompanying drawings and examples.

It should be noted that the following detailed description is illustrative and is intended to provide further explanation of the present application. Unless otherwise defined, all technical and scientific terms used herein have the same meanings commonly understood by one of ordinary skill in the art to which this application belongs.

Example 1

In this embodiment, a method for optimal dose selection of neonatal β-lactam drugs is disclosed, as shown in Figures 1 and 2, including:

S1: Obtain the newborn's characteristic information, selected drugs, and drug dosage and frequency information.

Characteristic information about the newborn includes demographic information and diagnostic information.

Demographic information included information such as gender, height, birth weight, current weight, BMI, gestational age, postpartum age, corrected gestational age, comorbid diseases, surgical history, and biochemical information.

Biochemical information includes renal function information, liver function information, and blood routine information.

Renal function information includes creatinine, uric acid, urea nitrogen, cystatin-C and carbon dioxide information; liver function information includes aspartate aminotransferase, alanine aminotransferase, total bilirubin, direct bilirubin, white blood Protein and globulin information; routine blood information includes platelet count, white blood cell count, red blood cell count, neutrophils, C-reactive protein and erythrocyte sedimentation rate information.

According to the diagnostic information of the newborn, the doctor selects a drug based on the diagnostic information. This embodiment is aimed at neonatal sepsis. Therefore, the drugs that the doctor can choose for neonatal sepsis are cefotaxime, ceftazidime, meropenem, and laxamethasone. Or commonly used beta-lactam drugs such as amoxicillin.

Based on the selected drug, the dosage and frequency information of the selected drug will be initially set.

S2: Based on the newborn's characteristic information, selected drugs, drug dosage and frequency information, and the trained drug optimal dose selection model, determine whether the drug dosage is the recommended optimal dose. Among them, the optimal drug dose selection The model takes the newborn's characteristic information, selected drugs, drug dosage and frequency information as input, and uses whether the drug dosage is the recommended optimal dose as the output, and is constructed through the CatBoost model.

The process of obtaining the trained optimal dose selection model for drugs is:

Obtain existing newborn characteristics information, drug selection information, dosage and frequency information, and adverse reactions as initial training data, and determine PD indicators under different drug dosages;

Label the initial training data according to the relationship between the PD index and the PD target as training data. When the PD index is greater than or equal to the PD target, the label of the initial training data is the recommended optimal dose. When the index is less than When targeting PD, the annotation label of the initial training data is not the recommended optimal dose;

The constructed optimal drug dosage selection model is trained with the training data, and the trained optimal drug dosage selection model is the trained optimal drug dosage selection model.

During the specific implementation, the initial training data obtained are all relevant data obtained with the written consent of the parents of the newborn, and the newborn's corrected gestational age is within 48 weeks, and the newborn has received other systemic experimental drug treatment, such as newborn ID , demographic information, diagnostic information, dosage and frequency information and other major research data are seriously missing, or other neonatal data that are not suitable for selection are deleted.

When determining the type of initial training data, data types with a missing rate greater than the set value in the acquired newborn data are deleted and used as the final initial training data.

In this embodiment, the initial training data is simulated through a population pharmacokinetic model to obtain the PD index of each drug under different dosages.

The population pharmacokinetic model uses a one-compartment or two-compartment model based on compartmental structure as the overall framework, equipped with an inter-individual variability model and a residual model. Among them, the inter-individual variability model is a power exponential method, and the residual model is Additive or proportional or mixed.

In order to improve the accuracy of the obtained PD indicators, existing population pharmacokinetic models were screened, and the optimal population pharmacokinetic model was selected as the population pharmacokinetic model for data simulation in this embodiment.

When simulating the initial training data through the population pharmacokinetic model, using different selected drugs and different drug dosages as simulation conditions, simulate the initial training data to obtain different drug concentrations under different dosages. PD indicator.

The drug concentration in the blood was obtained when different PD indicators were 50% dosing interval, 70% dosing interval and 100% dosing interval.

Any one of the three PD indicators can be selected for subsequent comparison with the PD target.

At present, there is no clear international consensus on which PD indicator is better. Therefore, this embodiment obtains the three most common PD indicators of the drug. According to actual needs, appropriate PD indicators can be selected and subsequently compared with the PD target to determine the optimal dose of the drug to improve the method disclosed in this embodiment. adaptability.

The selection criteria for PD targets are: taking the maximum MIC value of all pathogenic bacteria in sepsis as the PD target.

Summarize statistics on all pathogenic bacteria of neonatal sepsis at home and abroad, and calculate the MIC corresponding to all pathogenic bacteria under different drugs. For each drug, select an MIC that can cover all common pathogenic bacteria as the PD target. The optimal dose of the drug selected by this PD target is effective against all pathogenic bacteria of neonatal sepsis.

The initial training data and PD indicator data obtained through simulation and simulation under different conditions are summarized and integrated to obtain a newborn virtual database, which is used as training data for subsequent machine learning analysis.

Taking the newborn's characteristic information, selected drugs, drug dosage and frequency information as input, and using whether the drug dosage is the recommended optimal dose as the output, the CatBoost model is used to build the optimal dosage selection model of the drug. In addition, the optimal dosage of the drug is The optimal dose selection model can also output the adverse reactions corresponding to the administered dose.

The training data is randomly divided into a training set and a test set, and the constructed optimal dose selection model of the constructed drug is trained. After the training is completed, the trained optimal dosage selection model of the constructed drug is obtained.

The CatBoost model is not only good at processing classification features, but also can handle prediction offsets, thereby reducing the possibility of model overfitting; the built-in model of the CatBoost algorithm is a symmetric tree, which has fast training speed and is not prone to overfitting. This model can Excellent analysis and modeling of beta-lactam drug data for neonatal patients.

Input the newborn's characteristic information, selected drugs, drug dosage and frequency information obtained by S1 into the trained drug optimal dose selection model to determine whether the drug dosage is the recommended optimal dose.

When the drug dosage is not the recommended optimal dose, the drug dosage is readjusted so that the adjusted drug dosage is determined to be the recommended optimal dose through the trained drug optimal dose selection model.

In order to obtain the optimal dose of the drug, different drug dosages can be set, and then the recommended optimal dose can be screened through the trained drug optimal dose selection model, and the minimum value can be selected from the recommended optimal dose as The optimal dose of the drug.

Although the optimal dose of the drug and the corresponding adverse reactions are obtained in this embodiment, in specific applications, the optimal dose of the drug is only for doctors' reference and is not directly used for neonatal treatment.

In this embodiment, when predicting the optimal dose of a drug, the characteristic information of the newborn and the selected drug are comprehensively considered, and the drug dosage is determined by setting the relationship between the PD index and the PD target of the drug under the dosage. Whether the drug dose is the recommended best dose, and then select the minimum value from the recommended best dose as the best dose of the drug, achieving accurate prediction of the best dose of the drug; when selecting the PD target, all pathogenic bacteria of sepsis The MIC is taken into account so that the optimal dose can be selected to be effective against all pathogenic bacteria of sepsis.

This embodiment only realizes the prediction of the recommended optimal dose of the drug through the optimal drug dose selection model, and improves the prediction efficiency on the basis of ensuring the accuracy of optimal dose prediction.

Example 2

In this embodiment, an optimal dose selection system for neonatal β-lactam drugs is disclosed, including:

The data acquisition module is used to obtain the characteristic information of newborns, selected drugs, and drug dosage and frequency information;

The recommended optimal dose determination module is used to determine whether the drug dosage is the optimal dose based on the newborn's characteristic information, selected drugs, drug dosage and frequency information, and the trained drug optimal dose selection model, where , the optimal drug dose selection model takes the newborn's characteristic information, selected drugs, drug dosage and frequency information as input, and uses whether the drug dosage is the optimal dose as the output, which is obtained through the CatBoost model construction; when the drug is given When the drug dosage is not the recommended optimal dose, the drug dosage is readjusted so that the adjusted drug dosage is determined to be the recommended optimal dose through the optimal drug dosage selection model.

Example 3

In this embodiment, an electronic device is disclosed, including a memory, a processor, and computer instructions stored in the memory and executed on the processor. When the computer instructions are executed by the processor, a method disclosed in Embodiment 1 is completed. The steps described in the method for selecting the optimal dosage of beta-lactam drugs for neonates.

Example 4

In this embodiment, a computer-readable storage medium is disclosed for storing computer instructions. When the computer instructions are executed by a processor, the optimal dose selection of a neonatal beta-lactam drug disclosed in Embodiment 1 is completed. The steps described in the method.

Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those of ordinary skill in the art should understand that the present invention can still be modified. Modifications or equivalent substitutions may be made to the specific embodiments, and any modifications or equivalent substitutions that do not depart from the spirit and scope of the invention shall be covered by the scope of the claims of the invention.

Claims (8)

1. A method for selecting the optimal dose of neonatal β-lactam drugs, which is characterized by including: Obtain the newborn's characteristic information, selected drugs, and drug dosage and frequency information; Based on the newborn's characteristic information, selected drugs, drug dosage and frequency information, and the trained drug optimal dose selection model, determine whether the drug dosage is the recommended optimal dose. Among them, the drug optimal dose selection model, Taking the newborn's characteristic information, selected drugs, drug dosage and frequency information as input, and taking whether the drug dosage is the recommended optimal dose as the output, it is obtained through CatBoost model construction; When the drug dosage is not the recommended optimal dose, readjust the drug dosage so that the adjusted drug dosage is determined to be the recommended optimal dose through the optimal drug dosage selection model; The process of obtaining the trained optimal dose selection model for drugs is: Obtain existing newborn characteristics information, drug selection information, dosage and frequency information as initial training data, and determine PD indicators under different drug dosages; Label the initial training data according to the relationship between the PD index and the PD target as training data. When the PD index is greater than or equal to the PD target, the label of the initial training data is the recommended optimal dose. When the index is less than When targeting PD, the annotation label of the initial training data is not the recommended optimal dose; Use the training data to train the constructed optimal drug dosage selection model, and the trained optimal drug dosage selection model is the trained optimal drug dosage selection model; The maximum MIC value of all pathogenic bacteria in sepsis is regarded as the PD target. 2. A method for selecting the optimal dose of β-lactam drugs for newborns according to claim 1, characterized in that the minimum value is selected as the optimal dose of the drug from the recommended optimal doses. 3. A method for selecting the optimal dosage of β-lactam drugs for newborns according to claim 1, characterized in that the characteristic information of newborns includes demographic information and diagnostic information. 4. A method for optimal dose selection of neonatal β-lactam drugs as claimed in claim 1, characterized in that the initial training data is simulated through a population pharmacokinetic model to obtain different dosages of each drug. The PD indicator below. 5. A method for optimal dose selection of neonatal β-lactam drugs according to claim 4, characterized in that the population pharmacokinetic model uses different selected drugs and different drug dosages as simulation conditions. The initial training data is simulated to obtain the PD index of each drug under different dosages. 6. An optimal dose selection system for neonatal β-lactam drugs, which is characterized by including: The data acquisition module is used to obtain the characteristic information of newborns, selected drugs, and drug dosage and frequency information; The recommended optimal dose determination module is used to determine whether the drug dosage is the optimal dose based on the newborn's characteristic information, selected drugs, drug dosage and frequency information, and the trained drug optimal dose selection model, where , the optimal drug dose selection model takes the newborn's characteristic information, selected drugs, drug dosage and frequency information as input, and uses whether the drug dosage is the optimal dose as the output, which is obtained through the CatBoost model construction; when the drug is given When the drug dosage is not the recommended optimal dose, the drug dosage is readjusted so that the adjusted drug dosage is determined to be the recommended optimal dose through the optimal drug dosage selection model. 7. An electronic device, characterized in that it includes a memory and a processor and computer instructions stored in the memory and run on the processor. When the computer instructions are run by the processor, any one of claims 1-5 is completed. The steps of the method for selecting the optimal dose of neonatal β-lactam drugs. 8. A computer-readable storage medium, characterized in that it is used to store computer instructions. When the computer instructions are executed by a processor, the neonatal β-lactam drug of any one of claims 1-5 is completed. Steps in the Optimal Dose Selection Method.