Application of machine-learning models to predict the ganciclovir and valganciclovir exposure in children using a limited sampling strategy.

Intravenous ganciclovir and oral valganciclovir display significant variability in ganciclovir pharmacokinetics, particularly in children. Therapeutic drug monitoring currently relies on the area under the concentration-time (AUC). Machine-learning (ML) algorithms represent an interesting alternative to Maximum-a-Posteriori Bayesian-estimators for AUC estimation. The goal of our study was to develop and validate an ML-based limited sampling strategy (LSS) approach to determine ganciclovir AUC0-24 after administration of either intravenous ganciclovir or oral valganciclovir in children. Pharmacokinetic parameters from four published population pharmacokinetic models, in addition to the World Health Organization growth curve for children, were used in the mrgsolve R package to simulate 10,800 pharmacokinetic profiles of children. Different ML algorithms were trained to predict AUC0-24 based on different combinations of two or three samples. Performances were evaluated in a simulated test set and in an external data set of real patients. The best estimation performances in the test set were obtained with the Xgboost algorithm using a 2 and 6 hours post dose LSS for oral valganciclovir (relative mean prediction error [rMPE] = 0.4% and relative root mean square error [rRMSE] = 5.7%) and 0 and 2 hours post dose LSS for intravenous ganciclovir (rMPE = 0.9% and rRMSE = 12.4%). In the external data set, the performance based on these two sample LSS was acceptable: rMPE = 0.2% and rRMSE = 16.5% for valganciclovir and rMPE = -9.7% and rRMSE = 17.2% for intravenous ganciclovir. The Xgboost algorithm developed resulted in a clinically relevant individual estimation using only two blood samples. This will improve the implementation of AUC-targeted ganciclovir therapeutic drug monitoring in children.

Determining tissue distribution of the oral antileishmanial agent miltefosine: a physiologically-based pharmacokinetic modeling approach.

Miltefosine (MTS) is the only approved oral drug for treating leishmaniasis caused by intracellular Leishmania parasites that localize in macrophages of the liver, spleen, skin, bone marrow, and lymph nodes. MTS is extensively distributed in tissues and has prolonged elimination half-lives due to its high plasma protein binding, slow metabolic clearance, and minimal urinary excretion. Thus, understanding and predicting the tissue distribution of MTS help assess therapeutic and toxicologic outcomes of MTS, especially in special populations, e.g., pediatrics. In this study, a whole-body physiologically-based pharmacokinetic (PBPK) model of MTS was built on mice and extrapolated to rats and humans. MTS plasma and tissue concentration data obtained by intravenous and oral administration to mice were fitted simultaneously to estimate model parameters. The resulting high tissue-to-plasma partition coefficient values corroborate extensive distribution in all major organs except the bone marrow. Sensitivity analysis suggests that plasma exposure is most susceptible to changes in fraction unbound in plasma. The murine oral-PBPK model was further validated by assessing overlay of simulations with plasma and tissue profiles obtained from an independent study. Subsequently, the murine PBPK model was extrapolated to rats and humans based on species-specific physiological and drug-related parameters, as well as allometrically scaled parameters. Fold errors for pharmacokinetic parameters were within acceptable range in both extrapolated models, except for a slight underprediction in the human plasma exposure. These animal and human PBPK models are expected to provide reliable estimates of MTS tissue distribution and assist dose regimen optimization in special populations.

A machine learning approach to predict daptomycin exposure from two concentrations based on Monte Carlo simulations.

Daptomycin is a concentration-dependent lipopeptide antibiotic for which exposure/effect relationships have been shown. Machine learning (ML) algorithms, developed to predict the individual exposure to drugs, have shown very good performances in comparison to maximum a posteriori Bayesian estimation (MAP-BE). The aim of this work was to predict the area under the blood concentration curve (AUC) of daptomycin from two samples and a few covariates using XGBoost ML algorithm trained on Monte Carlo simulations. Five thousand one hundred fifty patients were simulated from two literature population pharmacokinetics models. Data from the first model were split into a training set (75%) and a testing set (25%). Four ML algorithms were built to learn AUC based on daptomycin blood concentration samples at pre-dose and 1 h post-dose. The XGBoost model (best ML algorithm) with the lowest root mean square error (RMSE) in a 10-fold cross-validation experiment was evaluated in both the test set and the simulations from the second population pharmacokinetic model (validation). The ML model based on the two concentrations, the differences between these concentrations, and five other covariates (sex, weight, daptomycin dose, creatinine clearance, and body temperature) yielded very good AUC estimation in the test (relative bias/RMSE = 0.43/7.69%) and validation sets (relative bias/RMSE = 4.61/6.63%). The XGBoost ML model developed allowed accurate estimation of daptomycin AUC using C0, C1h, and a few covariates and could be used for exposure estimation and dose adjustment. This ML approach can facilitate the conduct of future therapeutic drug monitoring (TDM) studies.

ChatGPT in pharmacometrics? Potential opportunities and limitations.

The potential of using ChatGPT in pharmacometrics was explored in this study, with a focus on developing a population pharmacokinetic (PK) model for standard half-life factor VIII. Our results demonstrated that ChatGPT can be utilized to accurately obtain typical PK parameters from literature, generate a population PK model in R and develop an interactive Shiny application to visualize the results. ChatGPT's language generation capabilities enabled the development of R codes with minimal programming knowledge and helped to identify as well fix errors in the code. While ChatGPT presents several advantages, such as its ability to streamline the development process, its use in pharmacometrics also has limitations and challenges, including the accuracy and reliability of AI-generated data, the lack of transparency and reproducibility regarding codes generated by ChatGPT. Overall, our study demonstrates the potential of using ChatGPT in pharmacometrics, but researchers must carefully evaluate its use for their specific needs.

Modelling the progression of illicit substance use patterns from real-world evidence.

AIMS: To investigate an innovative pharmacometrics approach that addresses the challenges of using real-world evidence to model the progression of illicit substance use. METHODS: The modelling strategy analysed real-world data from the National Longitudinal Study of Adolescent to Adult Health (AddHealth) survey using survival analyses and differential equations. Respondents were categorized into drug-naïve, active users and nonusers. The transitions between categories were modelled using interval-censored parametric survival analysis. The resulting hazard rate functions were used as time-dependent rate constants in a differential equation system. Covariate models for sex and depression status were assessed. RESULTS: AddHealth enrolled 6504 American teenagers (median age 16 years, range 11-21 years); this cohort was followed with five interviews over a 22-year period; the median age at the last interview was 38 years (range 34-45 years). The percentages of illicit drug users at Interviews 1-5 were 7.7%, 5.9%, 15.8%, 21.4% and 0.98%, respectively. The generalized gamma distribution emerged as the preferred model for the survival functions for transitions between categories. Age-dependent prevalence was obtained from the differential equation system. Active drug use was more prevalent in males, increased in adolescence and college years, peaked at 24 years, and decreased to low levels by 35 years. Depression, which was more frequent in females, increased the drug-naïve-active user transition rates but not the active user-nonuser and nonuser-active user transition rates. The evidence did not support an interaction between sex and depression. CONCLUSIONS: The model provided a satisfactory approximation for the age-dependent progression of illicit substance use from preadolescence to early middle age.

Population pharmacokinetic and dose-response models of nepadutant, a selective antagonist of the NK2 receptors, in infants with colic.

AIMS: The aim of the current analyses was to develop a population pharmacokinetic model for nepadutant in infants with colic, and a pharmacokinetic-pharmacodynamic model based on observations of duration of crying and fussing following oral nepadutant administration in infants (3-25 weeks) with colic. METHODS: The models were developed based on data obtained at baseline and following treatment with placebo, nepadutant 0.1 mg/kg or nepadutant 0.5 mg/kg administered for 7 days. A continuous response variable, duration of crying and fussing in minutes within 2 h interval, was assembled based on records from "baby's day" diary. RESULTS: The pharmacokinetics of nepadutant was described by a one-compartment model with first-order absorption and elimination with body weight as a structural covariate incorporated allometrically. For an infant weighing 5.3 kg, the estimated apparent clearance was 68.6 L/h (12% relative standard error) and exhibited large variability (78% coefficient of variation). The pharmacokinetic-pharmacodynamic model described (i) a circadian rhythm in the response with lowest and highest observations at 4 a.m. and 9 p.m., respectively, (ii) a placebo effect increasing and flattening out with time in an exponential manner, and (iii) a statistically significant (P < .01) linearly increasing response with dose. The observed and model predicted relative change in response from baseline was -35% and -28% (95% prediction interval -36%; -19%) following placebo, and -44% and -36% (-46%; -27%) after 0.5 mg/kg. CONCLUSIONS: Population pharmacokinetic and dose-response models were successfully developed characterizing the available nepadutant pharmacokinetics and duration of crying and fussing data in infants.

Population pharmacokinetic analysis of doravirine in real-world people with HIV.

AIMS: The pharmacokinetics of doravirine has been studied in clinical trials but not in real-world settings. Our study aims to characterize and identify factors influencing doravirine (a CYP3A4 substrate) pharmacokinetics in real-world people with HIV (PWH). METHODS: A total of 174 doravirine concentrations measured in 146 PWH followed up in the therapeutic drug monitoring (TDM) program at the University Hospital of Lausanne (Switzerland) between 2019 and 2023 were included in the analysis. Demographic data, clinical information and comedications were recorded during the routine SHCS visits (every 3-6 months). Population pharmacokinetic analysis and Monte Carlo simulations to investigate the clinical significance of the covariates retained in the final model were performed using NONMEM. RESULTS: A one-compartment model with first-order absorption and linear elimination best described doravirine pharmacokinetics. Potent CYP3A4 inhibitors and, to a lesser extent age, were the only tested covariates to significantly impact doravirine clearance (CL). Potent CYP3A4 inhibitors reduced CL by 50%, and a 30% decrease in CL was observed in an 80-year-old compared with a 55-year-old PWH. The effect of potent CYP3A4 inhibitors was prominent, explaining 59% of between-subject variability in CL. Model-based simulations predicted 2.8-fold and 1.6-fold increases in median steady-state trough and maximum doravirine concentrations, respectively, when a potent CYP3A4 inhibitor was co-administered. CONCLUSIONS: Our findings show that potent CYP3A4 inhibitors and age influence doravirine pharmacokinetics. However, given the good tolerability of doravirine, dosing adjustment of doravirine is probably not mandatory in those situations. TDM remains useful essentially in specific clinical situations, such as hepatic impairment, suspected nonadherence or pregnancy.

A physiological approach to renal clearance: From premature neonates to adults.

AIMS: We propose using glomerular filtration rate (GFR) as the physiological basis for distinguishing components of renal clearance. METHODS: Gentamicin, amikacin and vancomycin are thought to be predominantly excreted by the kidneys. A mixed-effects joint model of the pharmacokinetics of these drugs was developed, with a wide dispersion of weight, age and serum creatinine. A dataset created from 18 sources resulted in 27,338 drug concentrations from 9,901 patients. Body size and composition, maturation and renal function were used to describe differences in drug clearance and volume of distribution. RESULTS: This study demonstrates that GFR is a predictor of two distinct components of renal elimination clearance: (1) GFR clearance associated with normal GFR and (2) non-GFR clearance not associated with normal GFR. All three drugs had GFR clearance estimated as a drug-specific percentage of normal GFR (gentamicin 39%, amikacin 90% and vancomycin 57%). The total clearance (sum of GFR and non-GFR clearance), standardized to 70 kg total body mass, 176 cm, male, renal function 1, was 5.58 L/h (95% confidence interval [CI] 5.50-5.69) (gentamicin), 7.77 L/h (95% CI 7.26-8.19) (amikacin) and 4.70 L/h (95% CI 4.61-4.80) (vancomycin). CONCLUSIONS: GFR provides a physiological basis for renal drug elimination. It has been used to distinguish two elimination components. This physiological approach has been applied to describe clearance and volume of distribution from premature neonates to elderly adults with a wide dispersion of size, body composition and renal function. Dose individualization has been implemented using target concentration intervention.

Kinetic-pharmacodynamic model of warfarin for prothrombin time-international normalized ratio in Japanese patients.

AIMS: Genotype-guided dosing algorithms can explain about half of the interindividual variability in prothrombin time-international normalized ratio (PT-INR) under warfarin treatment. This study aimed to refine a published kinetic-pharmacodynamic model and guide warfarin dosage for an optimal PT-INR based on renal function. METHODS: Using a retrospective cohort of adult patients (>20 years) who were administered warfarin and underwent PT-INR measurements, we refined the kinetic-pharmacodynamic model with age and the genotypes of cytochrome P450 2C9 and vitamin K epoxide reductase complex subunit 1 using the PRIOR subroutine in the nonlinear-mixed-effect modelling programme. We searched the significant covariates for parameters, such as the dose rate for 50% inhibition of coagulation (EDR50 ), using a stepwise forward and backward method. Monte Carlo simulation determined a required daily dose of warfarin with a target range of PT-INR (2.0-3.0 or 1.6-2.6) based on the significant covariates. RESULTS: A total of 350 patients with 2762 PT-INR measurements were enrolled (estimated glomerular filtration rate [eGFR]: 47.5 [range: 2.6-199.0] mL/min/1.73 m2 ). The final kinetic-pharmacodynamic model showed that the EDR50 changed power functionally with body surface area, serum albumin level and eGFR. Monte Carlo simulation revealed that a lower daily dose of warfarin was required to attain the target PT-INR range as eGFR decreased. CONCLUSIONS: Model-informed precision dosing of warfarin is a valuable approach for estimating its dosage in patients with renal impairment.

Clinical antiviral efficacy of favipiravir in early COVID-19 (PLATCOV): an open-label, randomised, controlled, adaptive platform trial.

In early symptomatic COVID-19 treatment, high dose oral favipiravir did not accelerate viral clearance. BACKGROUND: Favipiravir, an anti-influenza drug, has in vitro antiviral activity against SARS-CoV-2. Clinical trial evidence to date is inconclusive. Favipiravir has been recommended for the treatment of COVID-19 in some countries. METHODS: In a multicentre open-label, randomised, controlled, adaptive platform trial, low-risk adult patients with early symptomatic COVID-19 were randomised to one of ten treatment arms including high dose oral favipiravir (3.6g on day 0 followed by 1.6g daily to complete 7 days treatment) or no study drug. The primary outcome was the rate of viral clearance (derived under a linear mixed-effects model from the daily log10 viral densities in standardised duplicate oropharyngeal swab eluates taken daily over 8 days [18 swabs per patient]), assessed in a modified intention-to-treat population (mITT). The safety population included all patients who received at least one dose of the allocated intervention. This ongoing adaptive platform trial was registered at ClinicalTrials.gov (NCT05041907) on 13/09/2021. RESULTS: In the final analysis, the mITT population contained data from 114 patients randomised to favipiravir and 126 patients randomised concurrently to no study drug. Under the linear mixed-effects model fitted to all oropharyngeal viral density estimates in the first 8 days from randomisation (4,318 swabs), there was no difference in the rate of viral clearance between patients given favipiravir and patients receiving no study drug; a -1% (95% credible interval: -14 to 14%) difference. High dose favipiravir was well-tolerated. INTERPRETATION: Favipiravir does not accelerate viral clearance in early symptomatic COVID-19. The viral clearance rate estimated from quantitative measurements of oropharyngeal eluate viral densities assesses the antiviral efficacy of drugs in vivo with comparatively few studied patients.

Synergizing pharmacometrics and pharmacovigilance for medication error management: the case of secukinumab.

PURPOSE: To demonstrate the effective integration of pharmacometrics and pharmacovigilance in managing medication errors, highlighted by a case involving secukinumab in a patient with hidradenitis suppurativa. METHODS: We present the case of a 41-year-old male with progressive hidradenitis suppurativa, unresponsive to multiple antibiotic regimens and infliximab treatment. Due to a medication error, the patient received 300 mg of secukinumab daily for 4 days instead of weekly, totaling 1200 mg. The regional pharmacovigilance center assessed potential toxicity, and a pharmacometric analysis using a population pharmacokinetic model was performed to inform dosing adjustments. RESULTS: Clinical data indicated that the received doses were within a non-toxic range. No adverse effects were observed. Pharmacometric simulations revealed a risk of underexposure due to the dosing error. Based on these simulations, it was recommended to restart monthly secukinumab injections on day 35 after the initial dose. Measured plasma concentrations before re-administration confirmed the model's accuracy. CONCLUSION: This case highlights the crucial collaboration between clinical services, pharmacovigilance, and pharmacometrics in managing medication errors. Such interdisciplinary efforts ensure therapeutic efficacy and patient safety by maintaining appropriate drug exposure levels.

Understanding time to peak effect of propofol as sole agent on bispectral index in children aged 2-12 years.

BACKGROUND: The pharmacodynamics of propofol in children have previously been described with the proprietary bispectral index (BIS) as an effect-site marker, and it has been suggested that the rate of onset of propofol might be age dependent, that is, a shorter time to peak effect in younger children. However, these analyses were potentially confounded by co-administered drugs, in particular opioids and benzodiazepines. Thus, the goal of this prospective study was to characterize the influence of age and weight on the onset of hypnotic effects from propofol, reflected by the time to peak of propofol effect-site concentration in the absence of additional drugs. METHODS: A total of 46 healthy children aged 2-12 years presenting for elective surgery were included in our observational cohort study. Solely propofol was administered via a target-controlled infusion pump programmed with the Paedfusor pharmacokinetic model. The BIS and infusion pump data were recorded. The effect of an induction "bolus" was recorded having stopped the pump once a propofol plasma target concentration of 7 µg.mL-1 was achieved. A direct-response and an indirect-response model in the context of nonlinear mixed-effects modeling was used to characterize and compare BIS data in children aged 2-6 years and older children aged 8-12 years. RESULTS: Time to peak of propofol effect-site concentration had a difference (p-value <.01) for age and weight, that is 84 [74, 96] (median [IQR] secs for children aged 2-6 years vs. 99 [91, 113] secs for children aged 8-12 years and 82 [71, 95] secs for weight 11-25 kg vs. 99 [91, 114] secs for weight 30-63 kg). The plasma effect-site equilibration rate constant for propofol had a heterogeneous distribution with a median of 2.36 (IQR: 2.05-2.93; range: 0.83-7.31) per minute but showed a weight-dependent effect in patients with weight below 45 kg. CONCLUSIONS: In children, the age and weight have an influence on time to peak effect of propofol. In the absence of opioids and benzodiazepines, time to peak effect was approximately 20% longer in children aged 8-12 years as compared to younger children. Such clinically relevant age and weight effects are an important consideration in the individualized titration of propofol dosing.

ChatGPT and Gemini large language models for pharmacometrics with NONMEM: comment.

This is a correspondence on "Evaluation of ChatGPT and Gemini large language models for pharmacometrics with NONMEM". Additional concern on using ChatGPT and Gemini is provided.

Training the next generation of pharmacometric modelers: a multisector perspective.

The current demand for pharmacometricians outmatches the supply provided by academic institutions and considerable investments are made to develop the competencies of these scientists on-the-job. Even with the observed increase in academic programs related to pharmacometrics, this need is unlikely to change in the foreseeable future, as the demand and scope of pharmacometrics applications keep expanding. Further, the field of pharmacometrics is changing. The field largely started when Lewis Sheiner and Stuart Beal published their seminal papers on population pharmacokinetics in the late 1970's and early 1980's and has continued to grow in impact and use since its inception. Physiological-based pharmacokinetics and systems pharmacology have grown rapidly in scope and impact in the last decade and machine learning is just on the horizon. While all these methodologies are categorized as pharmacometrics, no one person can be an expert in everything. So how do you train future pharmacometricians? Leading experts in academia, industry, contract research organizations, clinical medicine, and regulatory gave their opinions on how to best train future pharmacometricians. Their opinions were collected and synthesized to create some general recommendations.

Learning pharmacometric covariate model structures with symbolic regression networks.

Efficiently finding covariate model structures that minimize the need for random effects to describe pharmacological data is challenging. The standard approach focuses on identification of relevant covariates, and present methodology lacks tools for automatic identification of covariate model structures. Although neural networks could potentially be used to approximate covariate-parameter relationships, such approximations are not human-readable and come at the risk of poor generalizability due to high model complexity.In the present study, a novel methodology for the simultaneous selection of covariate model structure and optimization of its parameters is proposed. It is based on symbolic regression, posed as an optimization problem with a smooth loss function. This enables training of the model through back-propagation using efficient gradient computations.Feasibility and effectiveness are demonstrated by application to a clinical pharmacokinetic data set for propofol, containing infusion and blood sample time series from 1031 individuals. The resulting model is compared to a published state-of-the-art model for the same data set. Our methodology finds a covariate model structure and corresponding parameter values with a slightly better fit, while relying on notably fewer covariates than the state-of-the-art model. Unlike contemporary practice, finding the covariate model structure is achieved without an iterative procedure involving manual interactions.

Longitudinal modeling of efficacy response in patients with lupus nephritis receiving belimumab.

Belimumab was approved for active lupus nephritis (LN) in adults in the European Union and patients ≥ 5 years of age in the USA based on a Phase 3, double-blind, placebo-controlled, 104-week study. The study evaluated the efficacy of belimumab plus background standard therapy in adults with active LN using an intravenous (IV) dose of 10 mg/kg. A longitudinal analysis of Primary Efficacy Renal Response (PERR) and Complete Renal Response (CRR) was performed to assess whether patients with high proteinuria at the start of belimumab treatment would benefit from a higher dose. Responder probability was modeled as a logistic regression with probability a function of time and treatment (belimumab or placebo). Dropout risk at each visit was incorporated into a joint model of efficacy response; only efficacy data prior to dropout events (belimumab discontinuation, treatment failure, or withdrawal) were included. Average belimumab concentration over the first 4 and 12 weeks and baseline proteinuria were considered as continuous covariates. In general, renal response (PERR and CRR) over time was higher in patients receiving belimumab than in those receiving placebo. Baseline proteinuria was considered the most relevant predictor of renal response, with reduced efficacy in patients with increased proteinuria for both belimumab or placebo treatment. For belimumab-treated patients, belimumab exposure was not found to be an important predictor of renal response. In conclusion, the 10 mg/kg IV dose was considered appropriate in all patients and there was no evidence to suggest a higher response would be achieved by increasing the dose.

Mixed effect estimation in deep compartment models: Variational methods outperform first-order approximations.

This work focusses on extending the deep compartment model (DCM) framework to the estimation of mixed-effects. By introducing random effects, model predictions can be personalized based on drug measurements, enabling the testing of different treatment schedules on an individual basis. The performance of classical first-order (FO and FOCE) and machine learning based variational inference (VI) algorithms were compared in a simulation study. In VI, posterior distributions of the random variables are approximated using variational distributions whose parameters can be directly optimized. We found that variational approximations estimated using the path derivative gradient estimator version of VI were highly accurate. Models fit on the simulated data set using the FO and VI objective functions gave similar results, with accurate predictions of both the population parameters and covariate effects. Contrastingly, models fit using FOCE depicted erratic behaviour during optimization, and resulting parameter estimates were inaccurate. Finally, we compared the performance of the methods on two real-world data sets of haemophilia A patients who received standard half-life factor VIII concentrates during prophylactic and perioperative settings. Again, models fit using FO and VI depicted similar results, although some models fit using FO presented divergent results. Again, models fit using FOCE were unstable. In conclusion, we show that mixed-effects estimation using the DCM is feasible. VI performs conditional estimation, which might lead to more accurate results in more complex models compared to the FO method.

Low-dimensional neural ODEs and their application in pharmacokinetics.

Machine Learning (ML) is a fast-evolving field, integrated in many of today's scientific disciplines. With the recent development of neural ordinary differential equations (NODEs), ML provides a new tool to model dynamical systems in the field of pharmacology and pharmacometrics, such as pharmacokinetics (PK) or pharmacodynamics. The novel and conceptionally different approach of NODEs compared to classical PK modeling creates challenges but also provides opportunities for its application. In this manuscript, we introduce the functionality of NODEs and develop specific low-dimensional NODE structures based on PK principles. We discuss two challenges of NODEs, overfitting and extrapolation to unseen data, and provide practical solutions to these problems. We illustrate concept and application of our proposed low-dimensional NODE approach with several PK modeling examples, including multi-compartmental, target-mediated drug disposition, and delayed absorption behavior. In all investigated scenarios, the NODEs were able to describe the data well and simulate data for new subjects within the observed dosing range. Finally, we briefly demonstrate how NODEs can be combined with mechanistic models. This research work enhances understanding of how NODEs can be applied in PK analyses and illustrates the potential for NODEs in the field of pharmacology and pharmacometrics.

Evaluation of ChatGPT and Gemini large language models for pharmacometrics with NONMEM.

To assess ChatGPT 4.0 (ChatGPT) and Gemini Ultra 1.0 (Gemini) large language models on NONMEM coding tasks relevant to pharmacometrics and clinical pharmacology. ChatGPT and Gemini were assessed on tasks mimicking real-world applications of NONMEM. The tasks ranged from providing a curriculum for learning NONMEM, an overview of NONMEM code structure to generating code. Prompts in lay language to elicit NONMEM code for a linear pharmacokinetic (PK) model with oral administration and a more complex model with two parallel first-order absorption mechanisms were investigated. Reproducibility and the impact of "temperature" hyperparameter settings were assessed. The code was reviewed by two NONMEM experts. ChatGPT and Gemini provided NONMEM curriculum structures combining foundational knowledge with advanced concepts (e.g., covariate modeling and Bayesian approaches) and practical skills including NONMEM code structure and syntax. ChatGPT provided an informative summary of the NONMEM control stream structure and outlined the key NONMEM Translator (NM-TRAN) records needed. ChatGPT and Gemini were able to generate code blocks for the NONMEM control stream from the lay language prompts for the two coding tasks. The control streams contained focal structural and syntax errors that required revision before they could be executed without errors and warnings. The code output from ChatGPT and Gemini was not reproducible, and varying the temperature hyperparameter did not reduce the errors and omissions substantively. Large language models may be useful in pharmacometrics for efficiently generating an initial coding template for modeling projects. However, the output can contain errors and omissions that require correction.

Bayesian estimation in veterinary pharmacology: A conceptual and practical introduction.

Sophisticated mathematical and computational tools have become widespread and important in veterinary pharmacology. Although the theoretical basis and practical applications of these have been widely explored in the literature, statistical inference in the context of these models has received less attention. Optimization methods, often with frequentist statistical inference, have been predominant. In contrast, Bayesian statistics have not been widely applied, but offer both practical utility and arguably greater interpretability. Veterinary pharmacology applications are generally well supported by relevant prior information, from either existing substantive knowledge, or an understanding of study and model design. This facilitates practical implementation of Bayesian analyses that can take advantage of this knowledge. This essay will explore the specification of Bayesian models relevant to veterinary pharmacology, including demonstration of prior selection, and illustrate the capability of these models to generate practically useful statistics, including uncertainty statements, that are difficult or impossible to obtain otherwise. Case studies using simulated data will describe applications in clinical trials, pharmacodynamics, and pharmacokinetics, all including multilevel modeling. This content may serve as a suitable starting point for researchers in veterinary pharmacology and related disciplines considering Bayesian estimation for their applied work.