Making decisions during a pandemic is likely the most difficult task for governments nowadays. Those decisions are always backed by science, data, and the identification of vulnerabilities in each region and city. Therefore, technology has become an important and necessary ally during the pandemic.
To assist governments in the decision-making process in Colombia, a group of scientists, analysts and economists from the Colombian Institute for Health Technology Assessment (IETS, for its acronym in Spanish), proposed to the government in March to create an analytical tool for learning which places in Colombia are the most vulnerable in case of being infected with COVID-19.
In a few words, by crossing different data that the State has, it can be seen throughout the Colombian territory, block by block (or municipality in the case of smaller cities), what would be the possible health outcome in case of getting infected. Jhonathan Javier Rodríguez, an analytical expert, explained the use of this tool in a webinar organized by Afidro on June 24th.
The result of this interdisciplinary work is a “vulnerability viewer” that was formerly used on an exclusive basis by the government, but now it can be consulted by anyone. In addition, it is the only tool worldwide with a level of disaggregation block by block to understand vulnerability.
How does the viewer work?
The tool works with two variables: health and demography. The first, includes information of individuals with hypertension, diabetes, ischemic heart disease, chronic lung diseases, cancer, or immunodeficiencies in Colombia. To know which diseases could complicate a COVID-19 patient (over 80), healthcare professionals took part of the IETS group by making a content review.
The demographic variable indicates the location of +60-year old people, which households are overcrowded (more than three people per room), which households are intergenerational (people over 60 and people aged 20 to 29 in the same house) and the population density in different cities of Colombia.
With data of these two variables, the place of residence – according to the 2018 census – of people with chronic diseases was searched for and the percentage of these in each block of the entire territory was calculated. Based on this information, city blocks were divided into five different groups according to the level of vulnerability: low, medium-low, medium, medium-high and high. Machine learning was used to promptly achieve the results.
Then, using machine learning, it was possible to determine what is the maximum percentage of overcrowded households, intergenerational risks, and households with people over 60 years of age in each of the city blocks. The apples with the highest number of “highs” are the places with the highest vulnerability. This does not mean that there are more people infected in these sectors, it means that the health outcome may be worse.
More than vulnerability
The app also allows adding other variables to the vulnerability level such as multidimensional poverty, percentage of adults over 60, percentage of adults over 70, available medical centers and available hotels, since they can be used as support areas. In addition, it is possible to observe the mobility behavior of people. That is, if the most vulnerable have moved or stayed at home.
Using the vulnerability viewer, Colombian Government, through a Presidency-created website, it is possible to see the distribution of confirmed, recovered and deceased cases in all departments. In addition, it disaggregates information by age, gender and type (imported or related).
Bogota, la Guajira and El Chocó departments are using the information given by the viewfinder to make decisions regarding preventive and mandatory isolation. “If there is a person with COVID-19 in a high-risk area, it should preferably move to another area,” Rodríguez explained at the online event.
For example, in Bogotá, capital of Colombia, one of the most vulnerable sectors is Localidad de Usaquén, specifically in the Cedritos neighborhood. The above, according to Rodríguez, because there are many people over 60 years old with any health condition.
Development of analytical tools using machine learning during COVID-19 will contribute to a better outcome of the pandemic. These tools are very useful as they rapidly make estimates that human beings would take years to achieve. It is an opportunity to collect information and combine it with technology, in order to save as many lives as possible.