Identifying digital addiction—the compulsive use of the internet or social media—can be difficult for clinicians using only standard surveys. A large-scale review of 64 studies involving over 165,000 people found that machine learning algorithms are highly effective at identifying these behaviors. These tools showed a classification accuracy of 0.87 and an area under the curve (AUC) of 0.92, which indicates strong performance in distinguishing between healthy use and addiction.
The study looked specifically at internet and social media usage. It found that machine learning reached a 0.90 accuracy for internet addiction and 0.86 for social media addiction. Additionally, physiological markers showed high specificity of 0.90 when used to identify these conditions. These findings suggest that computer-based tools can provide reliable data for mental health professionals.
While these results are promising, it is important to remember that machine learning tools are not a replacement for a professional clinical diagnosis. The study also noted that more standardized diagnostic criteria and representative sampling are needed to improve accuracy further. For now, these tools serve as powerful screening instruments to help identify those who may need extra support.
