Project Manager/ Principal Investigator US FDA/CDER/OTS/OCS, Maryland
Drug safety is of great concern to public health. In addition, during the Investigational New Drug (IND) application submission process, the FDA specifically reviews the safety of the submitted drug candidate before the sponsor can initiate any clinical trials. SafetAI is a collaborative initiative led by suite of deep learning-based QSAR models for various safety endpoints critical to regulatory science and the IND review. We are developing a novel deep learning-based precision system for toxicity (DeepPST) which is designed to optimize toxicity prediction for individual compounds based on their chemical characteristics. In a pilot study, DeepPST was compared to several conventional machine learning and state-of-the-art deep learning methods for predicting drug-induced liver injury (DILI), carcinogenicity etc. The preliminary results from DeepPST yielded significant improvement in these toxicity endpoints in comparison to other deep learning and QSAR methods. DeepSafetAI facilitates drug-safety research with the novel DeepPST architecture that improves the “precision” in toxicity assessment by tailoring prediction to chemical characteristics. This presentation will describe deep learning model development for DILI using multiple types of standardized data1,2. We will also discuss ongoing validation process we adopted for testing these deep learning models to test its applicability in regulatory environment to check If these models could play a role in providing critical safety information.
Learning Objectives:
Understanding regulatory requirements: One potential learning objective for the SafetAI Initiative could be to understand the regulatory requirements for predicting toxicity endpoints and how AI-based predictive models can be used.
Improving regulatory decision-making: Another learning objective for the initiative could be to improve regulatory decision-making by providing more accurate and reliable data on toxicity endpoints.
Enhancing public health and safety: Ultimately, the goal of the SafetAI Initiative is to improve public health and safety by providing better tools for predicting and preventing toxic events.