Application of Machine Learning Classification to Improve the Performance of Vancomycin Therapeutic Drug Monitoring.

Bayesian therapeutic drug monitoring (TDM) software uses a reported pharmacokinetic (PK) model as prior information. Since its estimation is based on the Bayesian method, the estimation performance of TDM software can be improved using a PK model with characteristics similar to those of a patient. Therefore, we aimed to develop a classifier using machine learning (ML) to select a more suitable vancomycin PK model for TDM in a patient. In our study, nine vancomycin PK studies were selected, and a classifier was created to choose suitable models among them for patients. The classifier was trained using 900,000 virtual patients, and its performance was evaluated using 9000 and 4000 virtual patients for internal and external validation, respectively. The accuracy of the classifier ranged from 20.8% to 71.6% in the simulation scenarios. TDM using the ML classifier showed stable results compared with that using single models without the ML classifier. Based on these results, we have discussed further development of TDM using ML. In conclusion, we developed and evaluated a new method for selecting a PK model for TDM using ML. With more information, such as on additional PK model reporting and ML model improvement, this method can be further enhanced.

Development and Validation of Open-Source R Package HMCtdm for Therapeutic Drug Monitoring.

Most therapeutic drug monitoring (TDM) packages are based on the maximum a posteriori (MAP) estimation. In this study, HMCtdm, a new TDM package, was developed using a Hamiltonian Monte Carlo (HMC) simulation. The estimation process of HMCtdm for the drugs amikacin, vancomycin, theophylline, and phenytoin was based on the R package Torsten. The prior pharmacokinetic (PK) models of the drugs were derived from the Abbottbase® pharmacokinetics systems (PKS) program. The performance of HMCtdm for each drug was assessed through internal and external validations. The internal validation results of the HMCtdm were compared with those of a MAP-based estimation. The developed open-source HMCtdm package is user friendly. The validation results were reviewed and interpreted using the mean percentage error and root mean squared error. The successful transplantation of the prior PK structures (used in PKS) was confirmed by comparing the validation results with a MAP estimation. An open-source HMC-based TDM package was also successfully developed in this study, and its performance was evaluated. This package can be operated by users unfamiliar with C++ and can be further developed for various applications.