COLEV is a mixed-methods study on the design of Artificial Intelligence (AI) and data science-based solutions to inform public health responses to COVID-19 in different local health ecosystems. Our purpose is to produce and communicate evidence for differential public health measures to address COVID-19-related challenges in Colombia tailored to regional contexts and vulnerable populations, through the interdisciplinary rigorous ethical use of AI and data science (https://colev.uniandes.edu.co/en/).
The COLEV team developed evidence in different areas of decision-making and data by which it was possible to publish scientific articles and participate in over 10 academic events at the national and international levels, including:
Additionally, it was important for the COLEV team to work with a gender perspective. We had the support of Gender at Work and the Lady Smith Collective, with whom we worked actively in multiple sessions together with researchers from different parts of the global south. The main contributions of this collective materialized in the following results: 1. the identification of gender biases associated with the production of health indicators for the migrant population. 2. the production of an AI model for gender identification on Twitter as a first step analysis of the reactions to the accounts of the country's mayors.
Photo 1. TRACE and COLEV team
During the pandemic, the academy learned to reduce information analysis times and how to work collaboratively with other research centers, universities and decision-makers. We learned that communicating evidence requires taking into account several aspects such as:
Photo 2. Working on migrants and health data
The great data production of the last two years also leaves us with questions about the quality of the information we have and the biases associated with the way information is being generated and interpreted. It is necessary to improve the availability of datasets in the country and to generate mechanisms to facilitate navigating their diversity in terms of organizational ownership, sources of information, levels of disaggregation, and variables, among others. We identified a large number of databases and collection tools with incompatible formats and variables, a lack of capacity for uploading and sharing real-time information, and biases in data collection concerning populations, regions, and gender. Who is excluded from the data and what information associated with gender, ethnicity and class is visible or invisible in the databases we have today? How can we guarantee access to open data about the pandemic while complying with protection and security criteria? How can we develop skills and capacities for reading and analyzing data so that citizens continue to utilize health data?
Photo 3. Nairobi, Gender action learning
With the support of the IDRC and other organizations, a new research project was established. The new research proposal developed by Pontificia Universidad Javeriana and the Universidad de los Andes looks at how to adequately and effectively inform decision-making in response to infectious disease epidemics, and is currently being implemented. Furthermore, the learning experience of the COLEV gender group, as mentioned in the previous section, allowed the creation of a new gender group in a new research project: TRACE, including the participation of some members of the COLEV gender group in this new group.Tags: COVID-19, Mix-Methods, AI, Public Health