QSAR IN TOXICOLOGY .4. PREDICTION OF NONLETHAL MAMMALIAN TOXICOLOGICAL END-POINTS, AND EXPERT-SYSTEMS FOR TOXICITY PREDICTION

A critical review is given of quantitative and qualitative structure-a ctivity relationships (QSAR) for the prediction of some non-lethal mam malian toxicological endpoints required for regulatory purposes. QSAR in these areas of toxicology, such as irritation and respiratory aller gy, is hindered by a lack of quality in vivo toxicity data to model, a nd by the mechanisms of action not being fully understood. Much progre ss has been made though into the prediction of skin sensitisation. Com mercially available expert systems based on QSAR for toxicity predicti on are also described and evaluated critically. Generally it is consid ered although some systems are well advanced in particular areas, more work is required to make expert systems sufficiently accurate for rel iable toxicity prediction.