The proportion of electric shocks was as high as 59.6% because of component failure during use. In order to analyze the electric shock injury mode and assess the risk degree, this paper analyzed the transmission path of shock risk from the perspective of accidental release of electric energy, and combined entropy weight method and back propagation (BP) neural network method to build an electric appliances electric shock risk assessment model. 168 groups of sample data was used for training and simulation, the results prove that the model has a relative error percent of 2.80%, 4.97%, and 3.06% for low-risk, medium-risk, and high-risk evaluations, respectively. The assessment accuracy is high, and the evaluation results are close to the actual risk of the product. Through the comparative analysis of random sampling evaluation results, the accuracy of this model evaluation is improved compared with the traditional entropy weight method.