Design Of Individualized Dosing Regimen Based On Bayesian Inference

With the development of modern medical technology,a variety of medicines are being used in the clinical treatment of patients,significantly improving the cure rate and reducing post-operative mortality.However,the pharmacokinetic(PK)behavior of drugs is widely variable between individuals,so traditional dosing requires multiple blood samples to be taken from patients to adjust the dosage based on the doctor’s clinical experience,which does not meet the increasing clinical demands.Bayesian inference allows for the timely adjustment of dosing regimens by using limited blood samples to predict the expected level of treatment for patients.Based on this,this paper analyses the PK compartment model in therapeutic drug monitoring(TDM)by using a Bayesian inference approach to estimate individual PK parameters,and applies it to individualized vancomycin dosing regimen design to meet different clinical needs in TDM.Firstly,a PK compartment model for periodic dosing was proposed,in which the dosing period is fixed in clinical dosing and the drug exposure index under repeated dosing is related to the level of drug exposure in the prior period,while the trough concentration level during repeated dosing was used as the drug exposure index,which can better describe the changes in blood concentration during individual clinical dosing and assist in adjusting the dosing regimen.Secondly,to clarify the inter-individual variability in PK behavior,population PK analyses were performed,and a non-linear mixed-effects model was fitted to clinical back-testing of sparse datasets.For the Bayesian analysis of key PK parameters,a three-stage Bayesian hierarchical model was built,and two approaches,the empirical Bayesian and the fully Bayesian,were given for the estimation of individual PK parameters for each patient.Among which the fully Bayesian scheme uses the Monte Carlo Markov(MCMC)algorithm with the assistance of the Gibbs sampler or Hamiltonian Monte Carlo(HMC)sampler,which performs better in dealing with the parameter estimation problem of complex models.Finally,the established three-level Bayesian model and parameter estimation approach were applied to the vancomycin individual dosing regimen design.The improved PK model was used to calculate vancomycin exposure levels.Based on the typical values of population PK and individual PK parameter estimation,the initial dosing regimen,adjusted dosing regimen,and customized dosing regimen were designed,which met the needs of each stage of vancomycin clinical drug monitoring.At the same time,the individualized dosing decision system of vancomycin built in MATLAB platform gave the simulation of individual regimens and evaluated the prediction model.The results showed that the MCMC algorithm converged and the model fits well and had good prediction ability.The Bayesian-based individualized dosing scheme provides a more efficient and safe clinical dosing strategy with the help of mathematical models and statistical analysis approaches,and provides ideas for carrying out model-informed precision dosing.