March 23, 2023 — Researchers from the Barcelona Supercomputing Center-Centro Nacional de Supercomputación (BSC-CNS) will train a machine learning model to extract information from social media in real time and develop digital twins that will be used in the health crisis use case to inform reliable decision-making in crisis scenarios.

They will do this through the framework of CREXDATA (Critical Action Planning over EXtreme-Scale DATA), a Horizon Europe project that will push the frontiers of analytics, prediction, simulation and visualization to provide extremely precise, timely and useful information to support human and automated decision-making in critical situations. Through the envisioned prediction-as-a-service platform (PaaS), authorities will have access to the tools they need to confidently prepare for critical situations. CREXDATA launched on January 1, 2023. Its fifteen-partner consortium kicked-off the project in Athens, Greece.

CREXDATA will exploit extreme data to create simulation models and tools to mimic the properties of real-life extreme scale data streams from normal and critical conditions. These simulations and tools will be streamed into real-time predictive learning models that will be trained as information enters.

The resulting PaaS platform will allow action planners to easily register their various data stream sources and receive user-friendly predictive analysis workflows, supported by transparent AI techniques and Augmented Reality. These workflows will provide flexible, trustworthy and fit-for-purpose solutions that respond to users’ needs. BSC researchers will draw on their expertise in text mining to construct a multilingual language model that will be trained to operate online to extract crisis information from social media in real-time to support decision making.

Moreover, three use cases will evaluate CREXDATA technology: a weather emergencies use case, a health crisis use case and a maritime use case. BSC researchers will be responsible for the health crisis use case that will integrate current epidemiological and multi-scale simulation models with large-scale machine learning to allow for the development of flexible analytical platforms that support the decision-making process, including designing strategies for health crisis response and treatment optimization.

The development of a toolbox that will allow for the creation of digital twins addresses two scenarios within health crises: 1) tracking the movement and contagion of populations and 2) the treatment of patients. The envisioned predictive quality models allow for the quantification of the impact of an epidemic over time and patients’ responses to treatment, thus allowing for the development of appropriate containment measures for various scenarios.

This data-based approach requires the processing of extreme data to ensure that outbreaks are detected and tracked so that effective interventions can be planned.

BSC participation in these activities are being led by Arnau Montagud, Maite Melero and Miguel Ponce de Leon from the Life Sciences Department.

Alfonso Valencia, BSC principal investigator for this project stated, “Digital twins that mimic the behavior of infected populations or the comportment of a drug are crucial for harnessing data to ensure that decision-makers have the information they need to control and mitigate health crises.”

About CREXDATA

The CREXDATA (Critical Action Planning over EXtreme-Scale DATA) project is funded under Horizon Europe Research and Innovation Action number 101092749. This three-year project began on 1 January 2023. The Technical University of Crete leads this 15-partner consortium composed of: National Center for Scientific Research ‘DEMOKRITOS’, Universitaet Paderborn, RapidMiner GMBH, MarineTraffic, Barcelona Supercomputing Center, Fraunhofer Gesellschaft zur Foerderung der Angewandten Forschung E.V., Consiglio Nazionale delle Ricerche, Hydrometerological Innovative Solutions, Deutsches Rettungsrobotk-Zentrum e.V., Fire Department Dortmund/Stadt Dortmund, Universidad Rovira i Virgili, Finnish Meterological Institute, Disaster Competence Network Austria, Finland Ministry of Interior.

---

Source: BSC