New AI model can help patients with eating or drinking disorders

The model is trained to predict binge eating and drinking and could therefore help patients by alerting them via notifications. The new technology could contribute to a more efficient treatment of eating and drinking disorders.

New research by KU Leuven and the University of Berkeley, California gives new insights into the circumstances causing eating and drinking disorders. For a year, researchers observed the eating and drinking behaviour of 120 patients, as well as various emotional and environmental factors. Participants shared their experiences eight times a day, on certain days of the week. With all that data, through machine learning and artificial intelligence, researchers can now recognise patterns and predict binge eating and drinking. 

Some of the main predictors of binge eating or drinking, according to the study, are the time of the day, the social setting and various emotional factors. However, there are differences between patients who suffer from alcoholism and those who struggle with eating disorders. Positive emotions and social situations appear to be better predictors of binge drinking, while negative emotions are more closely related to binge eating.

The prediction model could help patients by indicating when eating or drinking binges are about to occur. Patients could record their emotions and behaviour via a smartphone, which the algorithm could then use to predict the risk of eating and drinking and send a notification when the risk is high.

"These insights are an important step forward in understanding and addressing the complexities in bulimia nervosa and alcohol use disorder," says KU Leuven researcher Nicolas Leenaerts.

© BELGA PHOTO SISKA GREMMELPREZ