 This work presents numerical simulations of a non-linear breathing transmission epidemic system using stochastic scale conjugate gradient neural networks, SCGGNNs, with loxagmoid activation function and 20 neurons in the hidden layers. The precision of SCGGNNs is obtained by comparing it to database solutions, and the results are presented using training, verification, and testing procedures to reduce mean-square error. Error histograms, regression values, correlation tests, and state transitions are used to approve the exactness and capability of the stochastic SCGGNNs. This article was offered by Naja Aboili, Muhammad Dalul Khan, and Sulkarnan Sabi.