About

In March 2020, The World Health Organization (WHO) announced the pandemic disease caused by SARS-CoV-2 (COVID-19) and brought measures in order to interrupt the spread of SARS-CoV2 worldwide. Understanding the transmission dynamics of infectious diseases in populations, regions and countries will contribute to better approaches to reducing the spread of these diseases.

COVIDAI aims to analyse many factors influencing COVID-19 development in humans, including genetical, blood markers, geographical position, as well as medical Roentgen images. His unique AI tool provides a set of solutions for COVID-19 development both in specific patients (personalised model), as well as on the level of region (epidemiological model).

COVIDAI tool would help medical experts to decide whether the patent will be subjected to further analysis and prescribe adequate therapy. Predictive models based on machine learning can provide useful data in terms of prediction of epidemiological events, which can save time for timely and optimal response of both the health system and the society.

 

Main activities performed within the COVIDAI project include development of two models:

  1. Personalized AI model for COVID-19 prediction (monitoring of patient’s condition and prediction of disease progress in time)
  2. Epidemiological model for COVID-19 (monitoring of number of people susceptible/exposed/infected/dead/recovered from COVID-19)

Personalized AI model for COVID-19 prediction

The developed disease progression tool uses machine learning methods to mine heterogeneous patient data provided by Clinical Centers in Rijeka, Croatia and Kragujevac, Serbia. The main aim of this tool is to assess the disease progression of the patient infected with COVID-19 in the next couple of days. This tool will work with the following data: demographic data; clinical image; blood test data, imaging data. The result of the model is prediction of the category risk of mortality.

A unique feature of COVID-19 interstitial pneumonia is an abrupt progression to respiratory failure. Our patient specific lung  models will focus on the  spread of virus-laden to  many regions of the lungs  from the initial site of  infection.

Epidemiological model for COVID-19

COVIDAI uses a compartmental epidemiological model, based on the partial differential equations   to describe the spread and clinical progression of COVID-19. The basic model structure is inspired by a number of studies on the natural clinical progression of COVID-19 infection. Based on the official statistical data for the countries Serbia and Croatia, we have calculated necessary parameters for our model, in order to predict COVID-19 spread in population. We can perform the same methodology for any other country.

Change over time the total number of infected globally. Comparison of actual data with the estimation obtained by genetic programming.

Change in time of the total number of deaths globally. Comparisons of actual data with estimation obtained by genetic programming.

Change in time of the total number of recovered patients globally. Comparisons of actual data with estimation obtained by genetic programming.

Epidemic curve for province Hubei (China). The graph represents the number of active cases through time. Comparisons of actual data with estimation obtained by genetic programming.

Global pandemic curve. The graph represents the number of active cases through time. Comparisons of actual data with estimation obtained by genetic programming.

Epidemic curve for Italy. The graph represents the number of active cases through time. Comparisons of actual data with estimation obtained by genetic programming.