Columbia Researchers Develop AI-Powered Smartphone App to Predict and Intervene in Teen Depression
Researchers at Columbia University, led by postdoctoral research scientist Isaac Treves, are developing a smartphone application that utilizes artificial intelligence (AI) to detect and intervene in depressive moods among teenagers. The study, conducted with adviser Randy P. Auerbach, found that AI's large language models can analyze smartphone data to predict the emergence of depressed moods within days. This approach differs from previous methods that relied on simpler, rule-based algorithms. The AI model, when fine-tuned by clinicians, agreed with human experts 84% of the time in labeling the sentiment of teen text entries. The proposed app aims to complement traditional therapy by identifying signals like negative self-talk and providing helpful nudges towards positive activities and thought patterns. It will feature two types of interventions: distraction and problem-solving, designed to engage teens in meaningful activities and offer concrete problem-solving techniques. The researchers are currently ...
