METHODS: There were 5 patients, with a median age of 1.75 (range 0.1-6.25) years, a median weight of 10.7 (2.9-21.5) kg, and a median creatinine clearance of 179 (44-384) mL/min/1.73m2, who received intravenous infusions of colistimethate each 8 hours. The median daily dose was 0.21 (0.20-0.21) million international units/kg, equivalent to 6.8 (6.5-6.9) mg of colistin base activity per kg/day. Plasma concentrations of colistimethate and formed colistin were subjected to population pharmacokinetic modeling to explore the patient factors influencing the concentration of colistin.

RESULTS: The median, average, steady-state plasma concentration of colistin (Css,avg) was 0.88 mg/L; individual values ranged widely (0.41-3.50 mg/L), even though all patients received the same body weight-based daily dose. Although the daily doses were ~33% above the upper limit of the FDA- and EMA-recommended dose range, only 2 patients achieved Css,avg ≥2mg/L; the remaining 3 patients had Css,avg <1mg/L. The pharmacokinetic covariate analysis revealed that clearances of colistimethate and colistin were related to creatinine clearance.

METHODS: A total of 28 critically ill patients were included in this study. All data were collected from medical, microbiology and pharmacokinetic records. The clinical response was evaluated on the basis of clinical and microbiological parameters. The 24-h area under the curve (AUC0-24) was estimated from a single trough level using established equations.

RESULTS: Out of the 28 patients, 46% were classified as responders to vancomycin treatment. The trough vancomycin concentration did not differ between the responders and non-responders (15.02 ± 6.16 and 14.83 ± 4.80 μg mL-1; P = 0.929). High vancomycin minimum inhibitory concentration (MIC) was observed among the non-responders (P = 0.007). The ratio between vancomycin trough concentration and vancomycin MIC was significantly lower in the non-responder group (8.76 ± 3.43 vs. 12.29 ± 4.85 μg mL-1; P = 0.034). The mean ratio of estimated AUC0-24 and vancomycin MIC was 313.78 ± 117.17 μg h mL-1 in the non-responder group and 464.44 ± 139.06 μg h mL-1 in the responder group (P = 0.004). AUC0-24/MIC of ≥ 400 μg h mL-1 was documented for 77% of the responders and 27% of the non-responders (c2 = 7.03; P = 0.008).

METHODS: C0 were retrieved from a large neonatal vancomycin dataset. Individual estimates of AUC0-24 were obtained from Bayesian post hoc estimation. Various ML algorithms were used for model building to C0 and AUC0-24. An external dataset was used for predictive performance evaluation.

RESULTS: Before starting treatment, C0 can be predicted a priori using the Catboost-based C0-ML model combined with dosing regimen and nine covariates. External validation results showed a 42.5% improvement in prediction accuracy by using the ML model compared with the population pharmacokinetic model. The virtual trial showed that using the ML optimized dose; 80.3% of the virtual neonates achieved the pharmacodynamic target (C0 in the range of 10-20 mg/L), much higher than the international standard dose (37.7-61.5%). Once therapeutic drug monitoring (TDM) measurements (C0) in patients have been obtained, AUC0-24 can be further predicted using the Catboost-based AUC-ML model combined with C0 and nine covariates. External validation results showed that the AUC-ML model can achieve an prediction accuracy of 80.3%.