We are a team of Spanish scientists who have been working since March 2020 in collaboration with the Valencian Government of Spain on using Data Science to help fight the SARS-CoV-2 pandemic. We have focused on 4 large areas of work: large-scale human mobility modeling via the analysis of aggregated, anonymized data derived from the mobile network infrastructure; computational epidemiological models; predictive models and citizen science by means of a large-scale citizen survey called COVID19impactsurvey which, with over 375,000 answers in Spain and around 150,000 answers from other countries is one of the largest COVID-19 citizen surveys to date. Our work has been awarded two competitive research grants.
Since March, we have been developing two types of traditional computational epidemiological models: a metapopulation compartmental SEIR model and an agent-based model. However, for this challenge, we opted for a deep learning-based approach, inspired by the model suggested by the challenge organizers. Such an approach would enable us to build a model within the time frame of the competition with two key properties: be applicable to a large number of regions and be able to automatically learn the impact of the Non-Pharmaceutical Interventions (NPIs) on the transmission rate of the disease.
This group is made up of more than twenty experts from the Universities and research centers of the Valencian Community (Spain) and led by Dr. Nuria Oliver. We have all been working intensively since the beginning of the pandemic, altruistically and using the resources available to us in our respective institutions and with the occasional philanthropic collaboration of some companies.
Affiliated with: Ellis Alicante, Universitat Jaume I, Universidad de Alicante, Universidad Miguel Hernández, Universitat Politècnica de València, Universidad Cardenal Herrera CEU.
We have developed machine learning-based predictive models of the number of hospitalizations and intensive care hospitalizations overall and for SARS-CoV-2 patients. We have also developed a model to infer the prevalence of the disease based on a few of the answers to our citizen survey https://covid19impactsurvey.org
We have developed 3 types of computational epidemiological models: (1) a metapopulation compartmental SEIR-type model; (2) an agent-based model and (3) a deep learning-based model
Our goal in the Prescription phase of the competition is to develop an interpretable, data-driven and flexible prescription framework that would be usable by non machine-learning experts, such as citizens and policy makers in the Valencian Government. Our design principles are therefore driven by developing interpretable and transparent models.
Given the intervention costs, it automatically generates up to 10 Pareto-optimal intervention plans. For each plan, it shows the resulting number of cases and overall stringency, its position on the Pareto front and the activation regime of each of the 12 types of interventions that are part of the plan.
POLITICO AI: Decoded: How AI is helping fight a pandemic — Europe’s coronavirus app — Insights from Valencia
XPRIZE And Cognizant Announce Grand Prize Winners In Pandemic Response Challenge To Restart Economies Around The World
XPRIZE and Cognizant Announce Grand Prize Winners in Pandemic Response Challenge to Restart Economies Around the World
Un trabajo sobre el coronavirus de Nuria Oliver y Alberto Conejero, premiado en Los Ángeles | Ciencia | Agencia EFE
Premiado un proyecto tecnológico español para ayudar a reactivar la economía durante la pandemia | Tecnología | EL PAÍS
Covid-19: La Comunidad Valenciana busca herramientas para afrontar otra ola de contagios | Comunidad Valenciana | EL PAÍS
Un modelo predictivo español coordinado por la ingeniera Nuria Oliver gana un desafío global para reactivar la economía | Economía
El equipo de Nuria Oliver gana el desafío "Respuesta a la pandemia" de XPRIZE para reactivar la economía
El sistema de inteligencia artificial de un equipo español gana el desafío internacional 'Respuesta a la pandemia'