Identifying environmental factors associated with tetrodotoxin contamination in bivalve mollusks using eXplainable AI

Area of Science:

• Marine biology
• Food safety
• Environmental science

Background:

• Tetrodotoxin (TTX) contamination in European bivalve mollusks poses food safety risks since 2012.
• Predicting TTX contamination is crucial for the food industry and regulatory bodies.
• Previous research identified shallow habitats and water temperature as potential drivers, but temporal relationships remain unclear.

Purpose of the Study:

• To develop and validate an explainable deep learning model for predicting TTX contamination in bivalve mollusks.
• To explore the temporal relationships between environmental factors and TTX contamination.
• To identify key environmental drivers of TTX contamination.

Main Methods:

• Development of an explainable, deep learning model.
• Input features included meteorological and hydrological data.
• Model output predicted the presence or absence of TTX contamination in the Dutch Zeeland estuary.

Main Results:

• The model identified photoperiod (sunrise/sunset times) and solar irradiance (global radiation) as significant predictors of TTX contamination.
• Water temperature and chloride concentration also contributed to TTX detection.
• The model demonstrated the influence of environmental factors on TTX presence.

Conclusions:

• Environmental factors, particularly photoperiod and solar irradiance, are associated with TTX contamination in bivalve mollusks.
• The findings support the hypothesis of an exogenous origin for TTX.
• The developed deep learning model serves as a valuable tool for mitigating marine toxin risks.