Identification of risk factors for antisocial behaviour by machine learning

Identification of risk factors for antisocial behaviour by machine learning

Over the last decades, scientists around the world have tried to explain the roots of antisocial behaviour and factors that might help predict it. Still, no upstanding answers have been provided. Based on ECPBHS data we wanted to identify those factors that would reveal future antisocial factors from childhood or youth and do to that, we used the machine learning method, which helped us to test more than 500 variables to find out whether any of them would be associated with antisocial behaviour.

We identified five factors that are associated with antisocial behaviour – substance use disorder, gender, aggressive behaviour, concentration difficulties and fighting in middle school. But we also found one protective factor – maternal education. Our findings have just recently published in the Psychiatry Research (https://www.sciencedirect.com/science/article/pii/S016517812030250X).