Probing Norwalk-like virus presence in shellfish, using artificial neural networks

Area of Science:

• Environmental microbiology
• Computational biology
• Food safety science

Background:

• Norwalk-like viruses (NLVs) are a significant cause of shellfish-borne illness.
• Accurate prediction of NLV presence in shellfish is crucial for public health.
• Traditional statistical models have limitations in predicting pathogen presence.

Purpose of the Study:

• To evaluate the efficacy of artificial neural network (ANN) models in predicting PCR-identified Norwalk-like virus presence/absence in shellfish.
• To compare the predictive performance of ANN models with logistic regression models.
• To assess the consistency of input variable importance across geographically diverse datasets.

Main Methods:

• Analysis of a shellfish database using artificial neural network (ANN) models.
• Comparison of ANN model performance with established logistic regression models.
• Separate analysis of two country-specific datasets using ANN models to assess geographical influence.

Main Results:

• ANN models demonstrated superior precision in predicting Norwalk-like virus presence and absence compared to logistic regression.
• Overall classification performance for ANN models reached 93%, significantly higher than the 75% achieved by logistic regression.
• ANN models effectively identified site-specific correlations between environmental indicators and pathogen presence.

Conclusions:

• Artificial neural network (ANN) modeling is a highly effective tool for predicting Norwalk-like virus contamination in shellfish.
• ANN models offer enhanced predictive accuracy and a deeper understanding of pathogen-indicator relationships compared to traditional methods.
• The findings support the application of ANN models for improved shellfish safety surveillance and risk assessment.