XTRACTIS For Formulation & Ecology
Prediction of the Toxicity of Chemical Molecule Residues & Discovery of New Nontoxic Herbicides
Design an AI-based decision system that efficiently, instantly, and rationally predicts the toxicity of residues of chemical molecules on animals.
Quickly discover the formulation of new molecules that have minimum impact on wildlife.
Help agrochemical manufacturers to measure the toxicity of existing molecules and check compliance with environmental regulations.
Help chemists design new bioactive molecules with less impact on small animals.
Reduce testing on laboratory animals.
- We get 6 decision systems; each system is composed of gradual rules without chaining, and each rule uses some of the variables that XTRACTIS identified as predictors.
- The most intelligible model uses 14 rules combining 29 predictors and the less intelligible one uses 11 rules sharing 39 predictors.
- Only a few rules are triggered at a time to compute the decision.
Overall, models have good Real Performances in Test.
It computes real-time predictions up to 70,000 decisions/second, offline or online (API).
XTRACTIS OPTIMIZE discovers a new molecule of herbicide with a global satisfaction degree of 0.995. This is quite impressive given the fact that the most nontoxic herbicide already known in the training dataset achieves an overall satisfaction degree of only 0.063!