A novel dose-response model for foodborne pathogens using neural networks

Foodborne infections are a significant cause of morbidity and mortality in human populations. Risk assessment and public health control measures could be greatly enhanced by establishing an accurate relationship between ingested dose and infection, and defining minimum infectious doses. In this paper, a novel neural network model is proposed for the dose response of foodborne pathogens. The proposed model assumes a three-layer structure with a fast backpropagation learning algorithm. The model predictions for four available data sets from the literature are compared using six statistical models (log-normal, log-logistic, simple exponential, flexible exponential, /spl beta/-Poisson and Weibull-Gamma). The methods of least square error, maximum likelihood and correlation coefficient are used for the comparison, and they show that the neural network model does better than the statistical models. Predictions of dose response for multiple types of pathogens and with different host age and gender using neural network models are discussed, with simulations.