Session 4. AI Advances in Computational Toxicology

The synergy of data generation and improvement of computers and algorithms has increased the power of AI more than billion-fold since AI was coined in 1956: Data in the world double every 18month, i.e., 90% of all date were produced in last three years; computer double in capacity every 24 months (Moore’s law) and AI algorithms double in capacity every 3 months since 2010. For most human skill tests, AI performs better than 90% of us and the latest LLM by ChatGPT achieved an IQ of 126 in the HaWi intelligence test.

For toxicology, AI promises support for data retrieval, evidence integration (systematic reviews, risk assessments), predictive toxicology of untested compounds, digital pathology and support in reporting. The prospects of animal replacement with better accuracy in (human) prediction, ethical benefits and cost-effectiveness are enormous. Beyond this, accelerated assessments with automated data analyses, real-time monitoring and complex analyses come into reach with user-friendly prediction tools. These changes also promise to democratize knowledge, encourage open-access databases, algorithms and publications. As a copilot for toxicology, it empowers researchers, regulators, consumers and industry. The development toward such a co-pilote in the EU ONTOX project is detailed here.

References

Hartung T. ToxAIcology -  AI as the New Frontier in Chemical Risk Assessment. Frontiers in AI, Sec. Medicine and Public Health 2023, 40:559–570. doi: 10.3389/frai.2023.1269932

Hartung T. ToxAIcology - the Evolving Role of Artificial Intelligence in Advancing Toxicology and Modernizing Regulatory Science. ALTEX 2023, 40: 559-570. doi: 10.14573/altex.2309191.

Kleinstreuer N and Hartung T. Artificial Intelligence (AI) – it’s the end of the tox as we know it (and I feel fine) - AI for Predictive Toxicology. Archives in Toxicology 2023, published online 20 Jan 2024. Doi: 10.1007/s00204-023-03666-2