Starting Year: 2020
Title: MATISSE: A machine learning based forecasting system for shellfish safety
Abstract: The marine environment provides a range of ecosystem services and benefits, including the
provision of protein food sources. Shellfish cultivation and harvesting from natural seed banks
respond to the increasing demand for seafood products and contributes to the economic
sustainability of coastal regions. However, shellfish may act as vectors of contaminants to
humans, and to safeguard public health, shellfish are routinely monitored for microbial quality,
metal contaminants, and marine toxins derived from a natural phenomenon named Harmful
Algal Blooms (HABs). A constant and statutory monitoring is in place to ensure that safe levels
are not exceeded. The presence of marine toxins is the most critical environmental factor
that affects shellfisheries, leading to recurrent closures to harvesting. The present strategy,
following the EU legislation, is reactive, thus able to respond only after shellfish contamination,
which frequently leads to severe economic losses and disruptions to the social fabric and
cultural identity of coastal towns. The MATISSE project aims at developing proactive strategies
to anticipate the environmental challenges posed to the shellfish industry. Based on a large
data set collected from several sources, including remote sensing products and historical data
from the routine environmental survey of the shellfish producing areas, useful forecasting
models will be built to guide management actions. Artificial intelligence and machine learning
tools will be developed to predict shellfish contamination based on the complex, high-
dimensional, time-series data provided by the different data sources. The predictive ability of
the models will be assessed through validation based on the historical in-situ measurements,
acquired via routine environmental surveys. The project have two main objectives: 1)
supporting the shellfish production sector through the prediction of toxins and faecal
contamination and anticipating changes to harvesting permissions, and 2) supporting the
Public Administration by improving the characterization of the shellfish production areas and
providing tools for a more adequate harvesting licensing and sustainable use of the coastline.
The project will provide a functional prototype to predict the risk of shellfish contamination,
which can be a powerful tool to anticipate closures, and mitigate economic losses.
Funding Source: FCT – Fundação para a Ciência e a Tecnologia, I.P.
Typology: R&D Project
Reference: DSAIPA/DS/0026/2019
Principal Investigator (PI): Marta B. Lopes
PI’s institution: NOVA LINCS
NOVA Math members involved: Marta B. Lopes