Method for calculating blood concentration of medicine

Abstract

The invention relates to a medicine blood concentration calculation method. The method comprises the following steps: S1, counting body information of a patient; s2, calculating the clearance rate of the medicine according to a formula estimated in S1; s3, according to a formula 3, estimating the apparent distribution volume of the patient; s4, according to a formula 4, calculating the change condition of the predicted value of the drug concentration in the body of the patient according to pharmacokinetics; s5, collecting the actual measured value of the drug concentration in the body of the patient, and recording the corresponding time and dosage; s6, defining a minimum damage function; s7, MCMC sampling is executed; s8, establishing a regression model by using a system learning method, constructing a personalized model of the in-vivo drug concentration calculation parameters of the patient, firstly constructing a (X, Y) data set, and obtaining Cl and Vd predictors of the patient; and S9, obtaining corresponding Cl and Vd by using the predictor obtained in the step S8, and obtaining a future in-vivo blood concentration change value of the patient to assist a doctor in medication.

Description

Technical Field

[0001] The present invention relates to the technical field of pharmaceutical monitoring, and more specifically to a method for calculating drug blood concentration. Background Art

[0002] In July 2022, the National Health Commission and the State Administration of Traditional Chinese Medicine jointly issued the "Notice on Further Strengthening Drug Safety Management and Improving the Level of Rational Drug Use" (hereinafter referred to as the "Notice"), requiring further strengthening of drug safety management, improving the level of rational drug use, and strengthening the use and management of key monitoring of rational drug use drugs, antimicrobial drugs, anti-tumor drugs, proton pump inhibitors, glucocorticoids, narcotics, Chinese medicine injections, etc. Through blood drug concentration monitoring, genetic testing, etc., identify drug risks, formulate individualized drug plans, optimize drug variety selection, and accurately determine drug dosages.

[0003] There are significant differences in pharmacokinetics due to differences in gender, age, comorbidities, concomitant medications, and genetic characteristics. Special populations such as infants, pregnant women, the elderly, and those with impaired liver and kidney function also have significant differences in the in vivo processes of drugs. There are also differences in the same individual with different dosage forms (such as injections, oral preparations, etc.) or different physiological states. Blood drug concentration monitoring (TDM) can effectively reflect the situation of drugs in the body and control drug concentrations within the therapeutic range, thereby reducing toxic side effects and improving drug efficacy.

[0004] Drug concentration monitoring (TDM) is guided by the pharmacokinetic principle and measures the concentration of drugs in the blood through chromatography or immunoassay to evaluate the efficacy or determine the dosing regimen. It is currently widely used to detect various clinical drugs with large side effects and narrow therapeutic range (such as antibacterial drugs, anti-epileptic drugs, immunosuppressants, anti-tumor drugs, etc.).

[0005] Some drugs have a narrow therapeutic window, and their therapeutic concentration is close to the toxic concentration, which is very easy to cause poisoning. Some drugs are difficult to obtain controllable blood concentrations. For example, after a certain dose of phenytoin sodium is used, the blood concentration increases sharply and nonlinearly, which poses a risk of poisoning. Long-term use of phenobarbital can easily lead to weakened body responsiveness and reduced efficacy. The original therapeutic effect must be achieved by gradually increasing the dose. Therefore, these drugs need to be adjusted in combination with blood concentration monitoring.

[0006] At present, only some departments in domestic tertiary hospitals monitor the blood drug concentration of some drugs, and the Cockcroft-Gault formula currently used is used to monitor and predict the blood drug concentration, but the parameters in the formula are group average values, which are not suitable for patients with different disease groups. Willems, JM found that when the patient is over 70 years old, the Cockcroft-Gault method is not as good as MDRD and CKD-EPI. Pierrat A also proved that Cockcroft-Gault is not used for patients under 12 years old. According to experimental results, based on the comparison of foreign open source mimic database and domestic hospital patient data, the error of Cockcroft-Gault in predicting the blood drug concentration of domestic patients is larger than that of foreign patients, and the mean absolute percentage error (MAPE) distribution is 0.47. And 0.41, indicating that there is a problem of large error when directly used for Chinese patients. The present invention proposes a new solution to the above problems. Summary of the invention

[0007] In view of the problems existing in the prior art, the purpose of the present invention is to provide a method for calculating the blood drug concentration of a drug to solve the technical problems mentioned in the background technology.

[0008] To solve the above problems, the present invention adopts the following technical solutions.

[0009] A method for calculating drug blood concentration, comprising:

[0010] S1: Collect the patient's physical information, including age, height, weight, gender, and serum creatinine concentration, where A represents age, H represents height, W represents weight, G represents gender, and Scr represents serum creatinine concentration. Use the Cockcroft-Gault formula to predict creatinine clearance based on A, H, W, G, and Scr:

[0011] Ccr=(140-A)×W×G / (Scr×72) (Formula 1)

[0012] In the formula, Ccr represents the predicted creatinine clearance of the patient, the unit of Ccr is mL / min, the unit of W is Kg, when the patient is male, G is 1, and when the patient is female, G is 0.85;

[0013] S2: Calculate the drug clearance rate (Clearance) according to the formula based on the Ccr estimated by S1.

[0014] Cl=k×Ccr+4 (Formula 2)

[0015] Where Cl stands for Clearance, its unit is the same as Ccr, and k is the estimated clearance of the drug;

[0016] S3: According to formula 3, estimate the patient's apparent distribution volume V based on A and W d

[0017] V d =0.29×A+0.33×W+11 (Formula 3);

[0018] S4: According to formula 4, the predicted drug concentration in the patient is calculated based on pharmacokinetics (C 预测 ) changes,

[0019]

[0020] Where C0 is the initial drug concentration value in the patient's body;

[0021] S5: Collect actual measurements of drug concentration in the patient C 实际 , and record the corresponding time and dosage;

[0022] S6: Define the minimum damage function loss,

[0023]

[0024] S7: Perform MCMC sampling to obtain the optimal Cl best and V d best;

[0025] S8: Use the system learning method to establish a regression model and build a personalized model for calculating the drug concentration parameters in patients. First, build the (X, Y) data set. X includes age, height, weight, creatinine value, heart rate, blood pressure, and body temperature. C 实际 Select the maximum value of the measurement point 24 hours before medication, and Y is the optimal value of the parameters Cl best and V at the corresponding measurement point d best, and Cl best and V d best constructs two different data sets; then the patient's Cl 预测 and V d预测 Based on the data set, regression analysis modeling is performed, and then the minimum loss value is selected as the final predictor, and finally the patient's Cl and V are obtained. d Predictor;

[0026] S9: Using the predictor obtained in S8, according to the basic information of the patient, the corresponding C 预测 and V d预测 , according to pharmacokinetic formula 4, calculate the patient's blood drug concentration and assist doctors in prescribing medication.

[0027] Furthermore, the MCMC is a method of approximate probability calculation, which uses sampling methods to reject and accept, and then obtains the posterior distribution of the parameters. The PyMc software package is used to build the model, and the likelihood function is constructed by measuring the value and the true value, and then MCMC is performed.

[0028] Furthermore, in the MCMC sampling process, the sampling times were 2000, the number of samples in the burn-in period was set to 1000, and the mean of the posterior distribution area of ​​the parameter was taken as the optimal value for each patient, that is, at each medication time point, the patient's optimal clearance Cl best and the apparent distribution container V d best dataset.

[0029] Furthermore, in the (X, Y) data set, X also includes whether the patient has sepsis, whether CRRT is used, albumin, urea nitrogen, and white blood cell characteristics.

[0030] Furthermore, the regression analysis adopts svm, lightGBM, and CNN modeling.

[0031] Compared with the prior art, the advantages of the present invention are:

[0032] 1. Based on the clinical data of Chinese patients, a model that conforms to Chinese physical signs was established, and the method proposed in this scheme has better prediction effect.

[0033] 2. Use more patient parameters and consider patient conditions more than traditional methods.

[0034] 3. This scheme proposes a standardized modeling method, which solves the problem of no training data in the past by constructing an intermediate variable data set and a prediction model. It is suitable for the calculation of blood drug concentrations of various drugs and has a wider range of uses. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 It is a system block diagram of the present invention. DETAILED DESCRIPTION

[0036] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention; it is obvious that the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments, and all other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making creative work are within the scope of protection of the present invention.

[0037] Example:

[0038] A method for calculating drug blood concentration.

[0039] S1: Collect the patient's physical information, including age, height, weight, gender, and serum creatinine concentration, where A represents age, H represents height, W represents weight, G represents gender, and Scr represents serum creatinine concentration. Use the Cockcroft-Gault formula to predict creatinine clearance based on A, H, W, G, and Scr:

[0040] Ccr=(140-A)×W×G / (Scr×72) (Formula 1)

[0041] In the formula, Ccr represents the predicted creatinine clearance of the patient, the unit of Ccr is mL / min, the unit of W is Kg, when the patient is male, G is 1, and when the patient is female, G is 0.85;

[0042] S2: Calculate the drug clearance rate (Clearance) according to the formula based on the Ccr estimated by S1.

[0043] Cl=k×Ccr+4 (Formula 2)

[0044] Where Cl stands for Clearance, its unit is the same as Ccr, and k is the estimated clearance of the drug;

[0045] S3: According to formula 3, estimate the patient's apparent distribution volume V based on A and W d

[0046] V d =0.29×A+0.33×W+11 (Formula 3);

[0047] S4: According to formula 4, the predicted drug concentration in the patient is calculated based on pharmacokinetics (C 预测 ) changes,

[0048]

[0049] Where C0 is the initial drug concentration value in the patient's body;

[0050] S5: Collect actual measurements of drug concentration in the patient C 实际 , and record the corresponding time and dosage;

[0051] S6: Define the minimum damage function loss,

[0052]

[0053] S7: Execute MCMC sampling to obtain the optimal Cl best and V dbest, MCMC is a method of approximate probability calculation. It rejects and accepts through sampling methods, and then obtains the posterior distribution of the parameters. The PyMc software package is used to build the model. The likelihood function is constructed by measuring the value and the true value, and then MCMC is performed. In the MCMC sampling process, the sampling times are 2000, and the number of samples in the burn-in period is set to 1000. The mean of the posterior distribution area of ​​the parameter is taken as the optimal value for each patient, that is, at each medication time point, the patient's optimal clearance Cl best and the apparent distribution container V d best dataset;

[0054] S8: Use the system learning method to establish a regression model and build a personalized model for calculating the drug concentration parameters in patients. First, build the (X, Y) data set. X includes age, height, weight, creatinine value, heart rate, blood pressure, body temperature, as well as whether the patient has sepsis, whether CRRT is used, and albumin characteristics. C 实际 Select the maximum value of the measurement point 24 hours before medication, and Y is the optimal value of the parameters Cl best and V at the corresponding measurement point d best, and Cl best and V d best constructs two different data sets; then the patient's Cl 预测 and V d预测 Based on the data set, regression analysis modeling was performed. SVM, lightGBM, and CNN were used for regression analysis. Then the minimum loss value was selected as the final predictor. Finally, the patient's Cl and V were obtained. d Predictor;

[0055] S9: Using the predictor obtained in S8, according to the basic information of the patient, the corresponding C 预测 and V d预测 , according to pharmacokinetic formula 4, calculate the patient's blood drug concentration and assist doctors in prescribing medication.

[0056] The above is only a preferred specific implementation of the present invention; however, the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical solution and its improved conception within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A method for calculating drug blood concentration, characterized in that: include: S1: Collect the patient's physical information, including age, height, weight, gender, and serum creatinine concentration, where A represents age, H represents height, W represents weight, G represents gender, and Scr represents serum creatinine concentration. Use the Cockcroft-Gault formula to predict creatinine clearance based on A, H, W, G, and Scr: Ccr=(140-A)×W×G / (Scr×72) (Formula 1) In the formula, Ccr represents the predicted creatinine clearance of the patient, the unit of Ccr is mL / min, the unit of W is Kg, when the patient is male, G is 1, and when the patient is female, G is 0.85; S2: Calculate the drug clearance rate (Clearance) according to the formula based on the Ccr estimated by S1. Cl=k×Ccr+4 (Formula 2) Where Cl stands for Clearance, its unit is the same as Ccr, and k is the estimated clearance of the drug; S3: According to formula 3, estimate the patient's apparent distribution volume V based on A and W d V d =0.29×A+0.33×W+11 (Formula 3); S4: According to formula 4, the predicted drug concentration in the patient is calculated based on pharmacokinetics (C 预测 ) changes, Where C0 is the initial drug concentration in the patient's body, C0 = dosage / V d , dosage is the dosage of the medication; S5: Collect actual measurements of drug concentration in the patient C 实际 , and record the corresponding time and dosage; S6: Define the minimum damage function loss, S7: Perform MCMC sampling to obtain the optimal Cl_best and V d _best; S8: Use the system learning method to establish a regression model and build a personalized model for calculating the drug concentration parameters in patients. First, build the (X, Y) data set. X includes age, height, weight, creatinine value, heart rate, blood pressure, and body temperature. C 实际 Select the maximum value of the measurement point 24 hours before medication, and Y is the optimal value of the parameters Cl_best and V at the corresponding measurement point d _best, and Cl_best and V d _best constructs two different data sets; then the Cl 预测 and V d预测 Based on the data set, regression analysis modeling is performed, and then the minimum loss value is selected as the final predictor, and finally the patient's Cl and V are obtained. d Predictor; S9: Using the predictor obtained in S8, the corresponding Cl is obtained according to the basic information of the patient. 预测 and V d预测 According to pharmacokinetic formula 4, calculate the patient's blood drug concentration C 预测 , Assist doctors in prescribing medicines. Assist doctors in prescribing medicines.

2. A method for calculating drug blood concentration according to claim 1, characterized in that: The MCMC is a method of approximate probability calculation, which uses sampling methods to reject and accept, and then obtains the posterior distribution of the parameters. The PyMc software package is used to build the model, and the likelihood function is constructed by measuring the value and the true value, and then MCMC is performed.

3. A method for calculating drug blood concentration according to claim 2, characterized in that: In the MCMC sampling process, the sampling times were 2000, the number of samples in the burn-in period was set to 1000, and the mean of the posterior distribution area of ​​the parameter was taken as the optimal value for each patient, that is, at each medication time point, the optimal clearance rate Cl_best of the patient was equal to the apparent distribution container V d _best dataset.

4. A method for calculating drug blood concentration according to claim 1, characterized in that: In the (X, Y) data set, X also includes whether the patient has sepsis, whether CRRT is used, albumin, urea nitrogen, and white blood cell characteristics.

5. A method for calculating drug blood concentration according to claim 4, characterized in that: The regression analysis adopts svm, lightGBM, and CNN modeling.

Images