Tacrolimus has a narrow therapeutic window and considerable variability in clinical use. Our goal was to compare the performance of multiple linear regression (MLR) and eight machine learning techniques in pharmacogenetic algorithm-based prediction of tacrolimus stable dose (TSD) in a large Chinese cohort. A total of 1,045 renal transplant patients were recruited, 80% of which were randomly selected as the “derivation cohort” to develop dose-prediction algorithm, while the remaining 20% constituted the “validation cohort” to test the final selected algorithm. MLR, artificial neural network (ANN), regression tree (RT), multivariate adaptive regression splines (MARS), boosted regression tree (BRT), support vector regression (SVR), random forest regression (RFR), lasso regression (LAR) and Bayesian additive regression trees (BART) were applied and their performances were compared in this work. Among all the machine learning models, RT performed best in both derivation [0.71 (0.67–0.76)] and validation cohorts [0.73 (0.63–0.82)]. In addition, the ideal rate of RT was 4% higher than that of MLR. To our knowledge, this is the first study to use machine learning models to predict TSD, which will further facilitate personalized medicine in tacrolimus administration in the future.

Linezolid is the first synthetic oxazolidone agent to treat infections caused by Gram-positive pathogens. Infected patients with liver dysfunction (LD) are more likely to suffer from adverse reactions, such as thrombocytopenia, when standard-dose linezolid is used than patients with LD who did not use linezolid. Currently, pharmacokinetics data of linezolid in patients with LD are limited. This study aimed to characterize pharmacokinetics parameters of linezolid in patients with LD, identify the factors influencing the pharmacokinetics, and propose an optimal dosage regimen. We conducted a prospective study and established a population pharmacokinetics model with the Phoenix NLME software. The final model was evaluated by goodness-of-fit plots, bootstrap analysis, and prediction corrected-visual predictive check. A total of 163 concentration samples from 45 patients with LD were adequately described by a one-compartment model with first-order elimination along with prothrombin activity (PTA) and creatinine clearance as significant covariates. Linezolid clearance (CL) was 2.68 liters/h (95% confidence interval [CI], 2.34 to 3.03 liters/h); the volume of distribution (V) was 58.34 liters (95% CI, 48.00 to 68.68 liters). Model-based simulation indicated that the conventional dose was at risk for overexposure in patients with LD or severe renal dysfunction; reduced dosage (300 mg/12 h) would be appropriate to achieve safe (minimum steady-state concentration [Cmin,ss] at 2 to 8 μg/ml) and effective targets (the ratio of area under the concentration-time curve from 0 to 24 h [AUC0–24] at steady state to MIC, 80 to 100). In addition, for patients with severe LD (PTA, ≤20%), the dosage (400 mg/24 h) was sufficient at an MIC of ≤2 μg/ml. This study recommended therapeutic drug monitoring for patients with LD. (This study has been registered in the Chinese Clinical Trial Registry under no. ChiCTR1900022118.)