CITIC presents the results of the use of artificial intelligence in the fight against COVID-19

CITIC, the University of A Coruña’s Centre for Information and Communications Technology Research and part of the CIGUS Network, today presented the results of two research projects based on Artificial Intelligence to combat the impact on health of COVID-10: the Data science and engineering programme for the evaluation and population and personalized prediction of the evolution of the COVID-19 disease (CEDCOVID), directed by Ricardo Cao Abad and Manuel F. González Penedo; and a study to predict the structure of SARS-CoV-2 proteins using Artificial Intelligence techniques, led by José Santos Reyes. Financed by the Galician Innovation Agency with a total of €336,895 from ERDF funds and the support of the First Vice-Presidency and the Galician Regional Ministry of Economy, Industry and Innovation, these projects valorise the practical ways in which Information and Communication Technologies (ICTs) can contribute to the integral improvement of health.

Mathematics at the service of healthcare administration

The research lines of the Data science and engineering project for the evaluation and population and personalized prediction of the evolution of the COVID-19 disease (CEDCOVID, IN845D2020/26) addressed a series of objectives, the most important of which are listed below:

The length of stay for hospitalisation was analysed, as well as the length of stay in the ICU of just over 10,000 Galician COVID-19 patients at the start of the pandemic, considering the influence of age, sex and a number of other clinical variables during these periods. The result was a model for predicting hospital and ICU congestion in the Galician health service for the various healthcare areas.

A web-based system for advanced data visualisation, extraction, transformation and uploading was implemented. Machine learning models were created to predict whether patients diagnosed with COVID require different levels of hospital care (ward admission or intensive care unit admission) during the course of their illness, based exclusively on demographic and clinical data. The effectiveness of non-pharmacological interventions was also studied in nine fields of activity to reduce SARS-CoV-2 transmission in Spain in the period between September 2020 and May 2021.

The impact of vaccination on COVID-19 incident rates in Spain was also analysed. For this purpose, statistical models were formulated to explain the number of hospitalised patients based on the cumulative incidence of COVID-19 and the percentage of the population vaccinated.

An Internet of Things infrastructure was implemented in which data is received from environmental sensors, industrial sensors and photovoltaic systems. Project actions also included the analysis and visualisation of data regarding CO2 levels in classrooms during the 2021 and 2022 University Admission Examinations (ABAU). Virtual simulations were conducted for places of special interest (such as residences, supermarkets, leisure or hospitality venues) in order to quantify the probability of contagion.

A model was developed for estimating the frequencies of SARS-CoV-2 virus variants present in wastewater samples based on the mutation frequencies of interest.

The search for new drugs based on knowledge of the three-dimensional structure of proteins

The aim of the project to predict the structure of SARS-CoV-2 proteins using AI techniques was to apply various artificial intelligence methods to predict the structure of the various protein components of SARS-CoV-2. Knowledge of the three-dimensional structure of proteins is crucial, as it is this structure that determines their function and interaction with other molecular components.

In this sense, efforts centred on the protein components whose structure could not be identified with standard laboratory methods. Computer prediction is necessary to identify the spatial position of each atom and amino acid in a specific protein, such as “non-structural” proteins involved in the virus replication process.

The project improved atomic resolution in various protein components with an initial structural model. The aim was to ensure maximum resolution and reliability in the three-dimensional protein structures for use in computational searches for potential drugs.

Artificial intelligence to combat Covid-19 – CITIC (udc.es)