Toxicity Prediction
Supporting medical decisions with explainable AI models that predict treatment toxicity while keeping clinicians in control.

- A growing number of restrictions on animal testing
- Important costs and delays
- High ethical and health risks
SubFrom mathematical optimization to operational reasoning.scribe to newsletter
Managing topical toxicological risk is always a priority
Everyday consumer products (cosmetics, household or ambient products), before they are placed on the market, must undergo a full safety assessment on all risks of human toxicity. When one of these consumer products comes into contact with the skin or eye, knowledge of the irritant and/or sensitizing potential of the ingredients that make up the product is necessary.
Historically, for skin sensitization, animal models have been used to determine both the hazard potential (is the product hazardous or not?) and the potency (i.e., the dose at which the product is hazardous) of the ingredients.
However, for several years now, manufacturers have been facing increasing ethical, scientific and regulatory pressures about this approach.
We looked for new approaches to risk studies, which will not use new data from animal or human tests, and which therefore take advantage of the development of alternative tests.
We aimed at combining IA paradigms with physico-chemical and toxicogenomics to focus on explainable predictions.
- Grouping molecules according to their toxicogenomic signatures
- Linking certain substructures to potential toxicity
- Determining important physicochemical properties, particularly related to dermal penetration
Let’s explore your problem together
Whether you're facing a known challenge or exploring what's possible, we're here to discuss how our approach can address your critical decision systems.
