Neural network modelling of the fate of Salmonella enterica serovar Enteritidis PT4 in home-made mayonnaise prepared with citric acid

Fifty-four mayonnaise recipes were generated by the central composite design and tested for microbiological safety at two temperatures (5 and 22 °C). The content of oil: (150–350 ml), egg yolk (10–35 g), citric acid (4.98% w/v) (10–40 g), salt (0–3 g), mustard (0–2 g), sugar (0–1 g) and white pepper (0.25 g) varied among the different recipes. The fate of Salmonella enterica serovar Enteritidis PT4 in mayonnaise products was investigated by both viable count and presence/absence tests and modelled by neural networks. This study demonstrated that feed-forward neural networks were incapable of modelling the survival/growth curves of S. Enteritidis PT4 as a one-step-procedure model, but were capable of modelling the presence/absence of the organism.