Computational Safety and Pharmacokinetic Assessment of Consumer Product APIs

← Case Studies

Case Study

Computational Safety and Pharmacokinetic Assessment of Two Consumer Product Active Pharmaceutical Ingredients (APIs)

CPI collaborated with a major consumer product company to apply advanced computational modelling, machine learning, and pharmacokinetic simulation approaches to support the safety and development assessment of two APIs. The project aligned with the companies' commitment to replace, reduce, and refine animal studies and, wherever possible, reduce the proportion of new product development involving animals.

• API safety
• Pharmacokinetics
• PBPK modeling
• Skin sensitization and irritation
• Off-target assessment

Challenge

The assessment focused on the safety and development of two active pharmaceutical ingredients, alongside dose-exposure relationships to support translation between clinical and preclinical settings.

What We Did

1. 1
   Acquired data, curated datasets, and created models

   CPI performed data acquisition, dataset curation, machine-learning model creation, model predictions, and PBPK model building.

2. 2
   Assessed safety and pharmacokinetic properties

   Predicted local and systemic safety characteristics of two known APIs and key pharmacokinetic parameters, including exposure and peak-concentration profiles, using established human data as a reference point.

3. 3
   Built and validated PBPK models

   Developed and validated PBPK models against published pharmacokinetic data. The models explored dose-exposure relationships across different physiological systems and assessed how closely predicted profiles aligned with established human data.

4. 4
   Explored skin sensitization and irritation modeling

   Examined publicly available safety datasets to assess the feasibility of predictive models. Read-across and machine-learning methods were used to analyze structurally related compounds and provide context for model predictions.

5. 5
   Assessed potential off-target interactions

   Identified potential biological interactions beyond the APIs' primary mechanisms of action, helping build a broader understanding of their safety and efficacy profiles.

6. 6
   Applied new modeling workflows and domain classification

   The work included new modeling workflows and CPI's recently published domain-classification approach.

Method reference: Harris, J. S.; Vignaux, P. A.; Lane, T. R.; Ekins, S. Uncertainty Quantification with Domain Classification Models for Acetylcholinesterase Inhibition. Journal of Chemical Information and Modeling 2026, 66 (16), 9829-9846. DOI: 10.1021/acs.jcim.6c00981 (accessed 8/27/2026).

What We Delivered

• Local and systemic safety predictions for two APIs.
• Dose-exposure relationships and predictions of exposure and peak-concentration profiles using established human data as a reference.
• PBPK models validated against published pharmacokinetic data.
• Read-across and machine-learning assessment of the feasibility of modeling skin sensitization and irritation.
• Potential off-target interactions to inform understanding of safety and efficacy profiles.
• Tailored machine-learning models using curated public toxicology datasets.

Conclusion

The results demonstrate how computational modelling can provide a robust framework for dose prediction, safety assessment, and translational research.

Assessing safety and pharmacokinetics while advancing non-animal approaches?

Talk To Our Team View Reference