Artificial intelligence enables faster detection of substance use disorder

A new artificial intelligence system developed by the University of Cincinnati can predict substance use disorder-defining behaviors with up to 83% accuracy and the severity of addiction with up to 84% accuracy.
The AI system uses a computational cognition framework, alongside traditional questionnaires and judgments about mildly emotional stimuli in the form of 48 pictures.
The study represents an innovative use of AI to assess and predict mental illnesses, potentially expanding to behavioral addictions such as social media, gaming, or excessive food consumption.
The researchers emphasize the system’s cost-effectiveness, scalability, and resilience to manipulation, potentially offering a low-cost, accessible first step for triage and assessment in mental health.
The identification accuracy of the substances used (e.g., stimulants, opioids, cannabis) reached up to 82%.
The AI system evaluated various behavioral economic concepts to formulate a profile of individuals with substance use disorder characterized by increased risk-seeking and reduced resilience to losses.
The study examined 3,476 participants, providing data for the machine learning algorithms to predict behaviors accurately.
The study was published in the Nature journal Mental Health Research and is proposed to enable assessment across a broader spectrum of addictions.