Introduction

The AI hackathon is a full-day immersive experience that serves as a replacement for our traditional student and postdoc workshop, offering an exciting opportunity to explore the intersection of artificial intelligence and sustainable practices. The AI Hackathon is designed to harness the potential of artificial intelligence (AI) in the field of toxicity. This one-day hackathon aims to bring together experts and enthusiasts on AI and green chemistry to develop predictive models for chemical toxicity using a comprehensive dataset. Traditional methods of toxicity testing are often time-consuming and expensive. AI offers a promising alternative, with the potential to revolutionize the way we approach chemical safety. Join us on this journey of predictive analytics for real-world sustainability challenges.

Tools and Technologies

Access link to the tutorials were shared with participants by e-mail. Participants can use their personal computers, Google Colab, or other processing platforms to build their models and get the final outcomes. 

• Challenges with the data
  Was the integration seamless? If not, how was it solved? Is the molecular representation reasonable?
• Good data practices
  Duplicate removal, data splitting, cross-validation, simple baselines.
• Understanding of the model
  The presenters can explain why they chose a certain model and how they approached hyper-parameter tuning.
• Discoveries from the data and modeling
  The presenters communicate their general insights from the hackathon. What did they learn from the model, for example what kind of molecular features lead to 'toxic' molecules? What can they say about the data reliability, e.g. was there enough data?
• Improvements
  What can be done to improve the results a next time. For example, using different models, different approaches to data cleaning etc.