To provide a method for using a neural network as a model for predicting the inflammation limit of a compound chemical mixture.

In the case of generating a neural network model by exercising data from critical variables of respective structure groups when the inflammation limit of the compound mixture is predicted by using the critical variables of the structure groups, the critical variables include compositional data and thermochemical data from the respective structure groups. This method includes testing the exercised data from the neural network model and verifying the exercised and tested data from the neural network to accurately predict the inflammation limit of a similar compound mixture having the same structure group.