The paper ‘COVID-BEHAVE dataset: measuring human behaviour during the COVID-19 pandemic’, published in Nature in December, was co-authored by UT scientists Kostas Konsolakis and professor Hermie Hermens from the Biomedical Signals and Systems group, alongside researchers from the University of Granada.
Konsolakis measured the behaviour of 21 participants from the Netherlands and Greece at the start of the pandemic in 2020. He was already planning to collect data for his PhD research, which focuses on measuring human behaviour via wearable solutions. Once the lockdown began, he saw it as a unique opportunity to examine human behaviour during a pandemic and a lockdown.
‘We measured physical, emotional, cognitive and social behaviour using sensor data from smartphones and activity trackers combined with questionnaires filled in by the participants,’ explains Konsolakis. ‘We collected data related to the number of steps and the intensity of the physical activities. For social behaviour, we measured the frequency of phone calls and text messages; while we used questionnaires to measure cognitive and emotional behaviour.’
‘Our dataset is unique,’ claims the PhD researcher. ‘It is not common to find a study with such a holistic approach. We focused on all four domains of human behaviour, while others usually only involve one. That is why we now have a unique dataset that provides insightful information about human behaviour.’
The main surprises? The involvement of the participants, says Konsolakis. ‘At the start, people were sceptical about participation. They were afraid of a breach of their privacy. However, this changed during the lockdown. People suddenly wanted to be involved, they wanted to contribute and help in a time of crisis. The volunteers were very committed and participated for about two months, some even longer.’
Changes in behaviour
When it comes to changes in people’s behaviour during the lockdown, Konsolakis is hesitant to make general conclusions. ‘It varied per person,’ he says. ‘Some people became less active during working hours, some became more engaged in cognitive tasks, while others not at all. We had participants from the Netherlands and from Greece, and so they were all in slightly different situations, but they were all affected by the lockdown.’
There were some common patterns, though. ‘Emotional behaviour was associated with social behaviour, for example. People felt happier when they engaged more with family and friends,’ says the UT scientist. ‘Moreover, people’s behaviour didn’t change from one day to the next. When rules changed, it took a while before people’s behaviour changed. It took time for people to adjust to both the lockdown and its end. In the future we could analyse this data in more detail to investigate changes based on different COVID-19 measures that were in place at the time.’
‘The main results tell us that it’s indeed possible to use recent technological advances, such as wearables and smartphones, to measure and possibly improve human behaviour,’ says Konsolakis. ‘These findings are important, because digital health is a big trend.’