Search

Summary: Background: Accurate prediction of the optimal dose for β-lactam antibiotics in neonatal sepsis is challenging. We aimed to evaluate whether a reliable clinical decision support system (CDSS) based on machine learning (ML) can assist clinicians in making optimal dose selections. Methods: Five β-lactam antibiotics (amoxicillin, ceftazidime, cefotaxime, meropenem and latamoxef), commonly used to treat neonatal sepsis, were selected. The CDSS was constructed by incorporating the drug, patient, dosage, pharmacodynamic, and microbiological factors. The CatBoost ML algorithm was used to build the CDSS. Real-world studies were used to evaluate the CDSS performance. Virtual trials were used to compare the CDSS-optimized doses with guideline-recommended doses. Findings: For a specific drug, by entering the patient characteristics and pharmacodynamic (PD) target (50%/70%/100% fraction of time that the free drug concentration is above the minimal inhibitory concentration [fT > MIC]), the CDSS can determine whether the planned dosing regimen will achieve the PD target and suggest an optimal dose. The prediction accuracy of all five drugs was >80.0% in the real-world validation. Compared with the PopPK model, the overall accuracy, precision, recall, and F1-Score improved by 10.7%, 22.1%, 64.2%, and 43.1%, respectively. Using the CDSS-optimized doses, the average probability of target concentration attainment increased by 58.2% compared to the guideline-recommended doses. Interpretation: An ML-based CDSS was successfully constructed to assist clinicians in selecting optimal β-lactam antibiotic doses. Funding: This work was supported by the National Natural Science Foundation of China; Distinguished Young and Middle-aged Scholar of Shandong University; National Key Research and Development Program of China.

The pharmacokinetic (PK) studies of meropenem in Chinese newborns with late-onset sepsis (LOS) are still lacking. Causative pathogens of LOS and their susceptibility patterns in China differ from the data abroad. We, therefore, conducted a developmental population pharmacokinetic–pharmacodynamic analysis in Chinese newborns with the goal to optimize meropenem dosing regimens for LOS therapy. An opportunistic sampling strategy was used to collect meropenem samples, followed by model building and validation. A Monte Carlo simulation was performed to show the probability of target attainment (PTA) for various dosages. The information from 78 newborns (postmenstrual age: 27.4–46.1 weeks) was compiled and had a good fit to a 1-compartment model that had first order elimination. The median (range) values of estimated weight–normalized volume of distribution (V)and clearance (CL) were 0.60 (0.51–0.69) L/kg and 0.16 (0.04–0.51) L/h/kg, respectively. Covariate analysis revealed that postnatal age (PNA), gestational age (GA) and current weight (CW) were the most important factors in describing meropenem PK. Simulation results showed for LOS with a minimal inhibitory concentration (MIC) of 8 mg/L, the doses of 30 mg/kg 3 times daily (TID) as a 1-h infusion for newborns with GA ≤ 37 weeks and 40 mg/kg TID as a 3-h infusion for those with GA > 37 weeks were optimal, with PTA of 71.71% and 75.08%, respectively. In conclusion, we proposed an evidence-based dosing regimen of meropenem for LOS in Chinese newborns by using the population pharmacokinetic–pharmacodynamic analysis, based on domestic common pathogens and their susceptibility patterns.

Objectives: Augmented renal clearance (ARC) of primarily renally eliminated antibacterial agents may result in subtherapeutic antibiotic concentrations and, as a consequence, worse clinical outcomes. Cefathiamidine is frequently used as empirical antimicrobial therapy in children with ARC, but pharmacokinetic studies in infants are lacking. This population pharmacokinetic study in infants with ARC was conducted to determine optimal dosing regimens of cefathiamidine.Methods: The population pharmacokinetics was conducted in 20 infants treated with cefathiamidine. Plasma samples of cefathiamidine were collected using opportunistic sampling, and the concentrations were detected by UPLC-MS/MS. Data analysis was performed to determine pharmacokinetic parameters and to characterize pharmacokinetic variability of cefathiamidine using nonlinear mixed effects modelling (NONMEM) software program.Results: The data (n = 36) from 20 infants (age range, 0.35–1.86 years) with ARC were fitted best with a 1-compartment model. Allometrically scaled weight and age as significant covariates influenced cefathiamidine pharmacokinetics. The median (range) values of estimated clearance and the volume of distribution were 0.22 (0.09–0.29) L/h/kg and 0.34 (0.24–0.41) L/kg, respectively. Monte Carlo simulations showed that the cefathiamidine doses of 100 mg/kg/day q12 h, 50 mg/kg/day q8 h and 75 mg/kg/day q6 h were chosen for bacteria with MIC 0.25, 0.5 and 2 mg/L, respectively.Conclusion: The population pharmacokinetic model of cefathiamidine for infants with ARC was developed. The PTA - based dosing regimens were recommended based on the final model.

ObjectiveCefepime is used to treat severe infections in neonates. Pharmacokinetic data have only been evaluated among preterm neonates and population pharmacokinetic model lacked external validation. Hence, our aim is to obtain the population pharmacokinetic parameters of cefepime with large sampling and optimize the cefepime dosage regimen for neonatal infection based on developmental pharmacokinetics–pharmacodynamics.MethodsBlood samples from neonates and young infants treated with cefepime were collected using the opportunistic sampling design. The concentration of cefepime was determined using high performance liquid chromatography with ultraviolet detection. The population pharmacokinetic model was established using NONMEM software.ResultsOne hundred blood samples from eighty-five neonates were analyzed. The population pharmacokinetics of cefepime were described by a one-compartment model with first-order elimination. Covariate analysis indicated that serum creatinine concentration, postmenstrual age and current weight had significant impact on the pharmacokinetic parameters of cefepime. Monte Carlo simulation results showed that the current dosage regimen (30 mg/kg, q12 h) had a high risk of insufficient dose. For 70% of neonates to obtain a higher free drug concentration than the minimum inhibitory concentration during 70% of the dosing interval, 50 mg/kg q12 h was needed with a susceptibility breakpoint of 4 mg/l. For a minimum inhibitory concentration of 8 mg/l, 40 mg/kg q8 h was recommended for all neonates.ConclusionA population pharmacokinetic model of cefepime in neonates and young infants was established. According to simulation results based on the developmental pharmacokinetics–pharmacodynamics, different dosage regimens should be given depending on pathogens and the postmenstrual age.