基于群体药动学模型的万古霉素个体化给药辅助决策平台的研制及应用

目的 研发一款适合临床实际应用场景的万古霉素个体化给药辅助决策平台，并为万古霉素的合理使用提供个体化给药建议。方法 基于当前已构建并验证可行的万古霉素群体药动学模型，运用Idea2019、JDK1.8、ETL等软件研发万古霉素个体化给药辅助决策平台。平台开发历经①需求分析；②模块设计；③软件测试、优化完善3个主要阶段。结果 成功研制并应用的万古霉素个体化给药辅助决策平台具有页面简约，功能完善，操作便捷的优势，按功能分为4个主要模块，分别是检索模块、信息模块、浓度预测模块以及报告模块。该平台可以连接医院内网平台，自动获取患者信息、用药信息、血药浓度检测结果等，并基于内嵌的群体药动学模型，结合患者个体信息，计算出个体药动学参数用于后续的万古霉素浓度预测。浓度预测模块将贝叶斯反馈法与患者用药信息、药物浓度测定结果及相关协变量参数值相结合，以指南推荐的谷浓度及AUC范围为目标值，计算满足目标浓度范围下的个体化给药方案，并考虑临床实际应用场景，设定自定义模拟功能，更具推广应用价值。结论 基于前期已构建的万古霉素群体药动学模型，在Idea2019、JDK1.8、ETL等软件工具辅助下，成功建立了万古霉素个体化给药辅助决策平台，该平台能更高效、便捷地辅助药师，为临床使用万古霉素提供个体化给药建议。

Development and Application of an Individualized Dosing-assisted Decision Platform for Vancomycin Based on Population Pharmacokinetic Model

OBJECTIVE To develop a vancomycin individualized dosing-assisted decision platform suitable for practical clinical application scenarios and provide individualized dosing recommendations for the rational use of vancomycin. METHODS Based on the vancomycin population pharmacokinetic model that had been constructed and verified to be feasible, the vancomycin individualized assisted decision-making platform was developed by using Idea2019, JDK1.8, ETL and other software tools. The platform development had gone through three main stages, included ①requirement analysis; ②design stage; ③software testing and optimization. RESULTS The vancomycin individualized assisted decision-making platform, which was successfully developed and applied, had the advantages of simple page, perfect function and convenient operation, and was divided into four main modules according to functions, namely retrieval module, information module, concentration prediction module and reporting module. The platform could connect to the hospital intranet platform to automatically obtain patient information, medication information and blood concentration test results, and calculate individual pharmacokinetic parameters for subsequent concentration prediction based on the embedded population pharmacokinetic model, combined with individual parameters. The concentration prediction module incorporated the Bayesian feedback method with patient medication information, drug concentration measurement results and relevant covariate parameter values, took the guideline-recommended trough concentration and AUC range as the target value, calculated the individualized drug administration scheme that met the target concentration range, and set up custom simulation functions considering the actual clinical application scenarios, which was of more popularization and application value. CONCLUSION Based on the vancomycin population pharmacokinetic model that has been successfully constructed in the previous stage, with the assistance of Idea2019, JDK1.8, ETL and other software tools, a vancomycin individualized dosing-assisted decision platform has successfully constructed, which can more efficiently and conveniently assist monitoring pharmacists to provide individualized dosing advice for clinical use of vancomycin.