Chilam – Laboratory for the Simulation of Complex Adaptive Systems

Introduction

The Covid-19 pandemic causes distress, human losses and economical crisis to all nations worldwide.
After a year of the first patient infected with the SARS.CoV-2 coronavirus, in Wuhan China, 137 million people got infected and almost 3 million died by Covid-19 disease.

Objective of the model

Can we predict the risks of Covid-19?
The Epidemiological Intelligence platform EpIPUMA can generate predictions trained on historical data that capture thousands of different conditions.

Predictive models

Predictive model on people ensembles

EpIPUMA predicts the personal risk of dying from Covid19, based on the vulnerability of a person (group age) and the risk of having Covid-19 within their district of residence. Vulnerability of having Covid-19 and the exposure to SARS-CoV-2 are combined in this model to predict the number of people that is likely to die in each district.

Map. Precision is defined here as the districts with the highest number of predicted deaths (highlighted in blue), that are confirmed by the number of people dead in the district (highlighted in red).

Machine learning

A complex system like the Covid-19 pandemic is adaptive and thus using machine learning is fundamental in EpIPUMA.
EpIPUMA learns and assess predictive relationships by considering all information available, for example demographics, mobility, and vulnerability.

Word cloud. Risk factors associated with finding Covid-19 cases in district without cases change continuously in time.