“Who Moved My Cheese”: Mouse vs. Machine-Based Prediction

SOT-AM19-Blog-Banner1.png

The Toxicological Sciences Featured Session on Tuesday during the SOT 58th Annual Meeting and ToxExpo titled, “Mouse vs. Machine … Are Animal Studies Being Supplanted by Computers?” debated the state of the art in computational toxicology. Like the famous parable in Spencer Johnson’s book Who Moved My Cheese?, adapting to change is difficult but necessary in life. For toxicologists, this change is exemplified by the application of computational toxicology to predict hazards of chemicals in lieu of in vivo testing.

Presenter Dr. Thomas Hartung likened artificial intelligence (AI) as the “new kid on the block” that can predict toxicity better than animal experiments. The importance of leveraging AI to reduce animal testing is well recognized. ECHA alone has data on more than 10,000 chemicals, including 800,000 studies, with hundreds of chemicals tested multiple times in the same in vivo assay. Dr. Hartung described a 2018 ToxSci publication by Luechtefeld et al. that was received in the news with great fanfare because AI was shown to outperform in vivo toxicity predictions (87% versus 81%, respectively).

Professor Ivan Rusyn cautioned attendees that there was no such thing as a free lunch and that better data beats fancier algorithms. A clear message from the session is that data curation is key to ensuring the accuracy of predictive models.

In 1785, Scottish poet Robert Burns wrote an ode to a mouse whose winter nest he had accidentally plowed over, which most of us remember as “the best laid schemes o’ mice and men often go astray.” Such a poem reminds us that the needed change from “mouse” to “machine” is unlikely to be perfect from the get-go. Embracing AI with a critical mindset and strict adherence to adequate data curation will minimize errors and improve predictability so that over time, AI predictions will match those predicted from in vivo testing.

This blog was prepared by an SOT Reporter. SOT Reporters are SOT members who volunteer to write about sessions and events they attend during the SOT Annual Meeting and ToxExpo. If you are interested in participating in the SOT Reporter program in the future, please email Giuliana Macaluso.

Recent Stories
SOT FDA Colloquium Will Explore 3D Bioprinted Tissue Models for Predictive Safety

Memoirs of an SOT Annual Meeting Veteran

Share in the Science Presented during the SOT 58th Annual Meeting and ToxExpo