Skip to main content
Contact Info
Swapnil Chavan

Swapnil Chavan got his masters in Pharmaco-informatics from NIPER, India, and received his  PhD in Computational Chemistry from Linnaeus University, Sweden. Then he went to NCCT  (National Center for Computational Toxicology) at US – Environmental Protection Agency for  his postdoctoral research where he served as a Unilever postdoc. Swapnil currently works as a  senior data scientist in computational toxicology at RISE AB, Sweden.  

INTERESTS: He leads innovative research towards developing novel chemical risk-assessment  methods using cheminformatics, bioinformatics, image-informatics and machine learning techniques. Currently, he is focusing on integrating explainable AI and uncertainty estimation  approaches towards building a comprehensive toxicity prediction tool. 

OpenTox 2023 Virtual Conference 

Session description: 

In recent years, artificial intelligence (AI) has revolutionized drug discovery research by significantly cutting down time as well as cost associated with multi-stage drug discovery process. AI has been successfully employed for various purposes in a drug discovery e.g. de-

novo design, target identification, drug property prediction, reaction prediction, toxicity screening, drug prioritization, etc. Despite the increasing success stories of AI applications, the underlying AI models often sound incomprehensible to interpretation by the human brain.  Therefore, there is a need for ‘explainable’ AI (XAI) approaches to make these black-box AI  models more transparent and interpretable. 

The biggest question AI/ML community facing today is which XAI approach to be used? -model-dependent OR model agnostic?  

-global OR local? 

-gradient-based OR surrogate model-based OR perturbation-based? 

Within OpenTox virtual conference 2023, we are aiming to discuss various XAI approaches and their practical applications in the field of predictive toxicology. Understanding how XAI can help in model de-bugging as well as understanding uncertainty within model’s prediction will be the second discussion point for this session.