English  Français

Using data science and machine learning techniques to understand behavioural responses to COVID-19 in Kenya

Francis Obare, Population Council
Louis Graham, Busara Center for Behavioral Economics
George Odwe, Population Council
Mathilde Schilling, Busara Center for Behavioral Economics
Daniel Mwanga, Population Council
Jaspreet Singh, Busara Center for Behavioral Economics
Eva Muluve, Population Council

The low adherence to COVID-19 prevention measures in some parts of Africa and the wide variations in vaccine availability and coverage underscore a need for an in-depth understanding of the behavioural responses to the prevention measures in different contexts. This study will examine the extent to which perceptions about COVID-19 and perceived vulnerability to infection influence adoption of protective behaviours in Kenya, variations across different sub-groups of people, and the implications of current patterns for future risks, needs and opportunities for the control of COVID-19 and similar pandemics. It will use data science and machine learning techniques to generate new insights from two existing datasets: 1) a longitudinal COVID-19 knowledge, attitude and practice (KAP) survey conducted in five informal settlements in Nairobi County, and in rural settings in Wajir, Kilifi and Kisumu Counties of Kenya among 3,661 adult women, 1,826 adult men, 2,825 female adolescents, and 1,106 male adolescents; and 2) data from Twitter platform on COVID-19 related messages posted by users in Kenya. Findings will inform targeted information, education and communication (IEC) as well as prevention programmes (including the roll-out of COVID-19 vaccines) to those most in need who might be overlooked by existing interventions.

Keywords: COVID-19, Big data / Social media, Longitudinal studies, Health and morbidity

See extended abstract.

  Presented in Session P14.