QSAR models are developed using a dataset of chemicals with known toxicological properties. Various molecular descriptors are calculated to represent the chemical structure. Statistical or machine learning methods are then applied to identify relationships between these descriptors and the observed toxicological effects. The resulting model can predict the toxicity of new chemicals based on their molecular structure.
