KU Leuven and UC Berkeley Research Unveils New Insights into Binge Eating and Drinking in Bulimia and Alcohol Use Disorders
Research from the Mind-Body Research group of KU Leuven and the University of California, Berkeley sheds new light on the conditions that cause binge eating and binge drinking in people with bulimia nervosa and alcohol use disorder. The groundbreaking research, published in Psychological Medicine , uses advanced techniques in machine learning and artificial intelligence to develop prediction models for binge eating and binge drinking in real-life. These offer promising insights and enable more effective treatments.
The study followed 120 patients over a period of 12 months. Their eating and drinking behavior was studied, as well as various emotional, behavioral and contextual factors. This was done using the so-called experience sampling method (ESM), in which participants reported their behavior and experiences eight times a day on specific days of the week.
This intensive and very extensive data collection then made it possible to search for patterns and predictors of problematic eating or drinking behavior. Machine learning models and artificial intelligence were used for this.
Group-level models were developed, as well as models tailored to individual patients. The researchers found that group-level models, which pool data from multiple patients, generally outperformed individual-level models. Some of the most important predictors of binge eating and binge drinking were the experience of an urge to eat or drink and the time of day. In addition, social context and emotional factors also played an important role.
Just in time support
The insights from this research enable just-in-time adaptive interventions (JITAI), which can provide support exactly when patients need it most.
For example, a patient could use a smartphone app to report their emotions and behavior, and an algorithm could predict their risk of binge eating or binge drinking. When the risk is high, an alert could be sent immediately, tailored to the situation the patient is in at that moment. Such a fine-grained approach could significantly improve the effectiveness of treatments for bulimia nervosa and alcohol use disorder.
This granularity is necessary, because the study showed that the social context and emotional state of patients differed significantly between binge eating and binge drinking. For example, positive emotions and a social situation appeared to be more predictive of binge drinking, while negative emotions were more closely linked to binge eating.
“Our research shows that understanding the context in which the behavior occurs is crucial,” Leenaerts adds. “This insight allows us to develop interventions that are not only timely, but also contextually relevant and take into account the emotional state of patients