Developing Predictive Models to Facilitate Interpretation of Toxicology Study Results
Project Description |
|---|
This project will evaluate and enhance the PHUSE computational pipeline designed to predict target organs of toxicity using SEND datasets. Team members will assess the pipeline’s feasibility and performance on internal organisational data, update it for broader compatibility with diverse database systems, and improve its robustness across heterogeneous data sources. The project will also explore additional study‑level interpretations – such as adversity, NOAEL determination, clinical translatability, and structure–activity relationships – to expand the range of predictive modelling approaches. Successful methods will be submitted for publication in peer‑reviewed journals. SEND has enabled the creation of large toxicology databases suitable for training models on expert interpretations of historical studies. These models can streamline future study review by predicting likely interpretations, reducing manual effort for both industry toxicologists and regulators. They may also highlight which endpoints are most informative for predicting specific toxicities, such as hepatotoxicity or nephrotoxicity. |
Project Leads | |
|---|---|
Kevin Snyder, Centara | |
Lennart Anger, Genentech | |
Alex Pearce, PHUSE Project Coordinator |
Click above for the latest project status. |