In this study, the researchers used player tracking data to evaluate whether behavioural indicators could predict problem gambling using machine learning models. The data were obtained from 1,743 adults from the UK, Canada, and Spain who played online casino games. The researchers found that behavioural indicators (e.g., frequent deposits within a session) were more important than monetary indicators (e.g., total amount of money gambled) in predicting problem gambling. The five machine learning models employed in this study showed promising results in predicting problem gambling. Including country-specific data improved their accuracy. Nevertheless, the models performed well even without training using country-specific data. The findings suggest that there are behavioural indicators that can be used across diverse contexts to identify problem gambling.
