The potential of using chatGPT in pharmacometrics was explored in this study, with a focus on developing a 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 identify and 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 interpretability of AI. Overall, our study demonstrates the potential of using chatGPT in pharmacometrics, but researchers must carefully evaluate its use for their specific needs.

Background Recombinant factor IX Fc fusion protein (rFIX-Fc) is an extended half-life (EHL) factor concentrate administered to haemophilia B patients. So far, a population pharmacokinetic (PK) model has only been published for patients ≥12 years of age. Aim Assess the predictive performance of the published rFIX-Fc population PK model for patients of all ages and develop a model that describes rFIX-Fc PK using real world data. Methods We collected prospective and retrospective data from patients with haemophilia B (FIX activity level ≤5 IU/dL) treated with rFIX-Fc and included in the OPTI-CLOT TARGET study (NTR7523) or United Kingdom (UK)-EHL Outcome Registry (NCT02938156). Predictive performance was assessed by comparing predicted with observed FIX activity levels. A novel population PK model was constructed using nonlinear mixed-effects modelling. Results Real world data was obtained from 37 patients (median age: 16 years, range 2-71) of whom 14 were <12 years of age. Observed FIX activity levels were significantly higher than levels predicted using the published model, with a median prediction error (PE) of -48.8%. The novel model showed a lower median PE (3.4%) and better described rFIX-Fc PK, especially for children <12 years of age. In the novel model, an increase in age was correlated with a decrease in clearance (p<0.01). Conclusion The published population PK model significantly underpredicted FIX activity levels. The novel model better describes rFIX-Fc PK, especially for children <12 years of age. This study underlines the necessity to strive for representative population PK models, thereby avoiding extrapolation outside the studied population.