eRisk 2025: contextual and conversational approaches for depression challenges

Since its foundation in 2017, the eRisk CLEF Lab has pioneered research in early risk detection on the Internet, focusing on various mental health challenges such as depression, anorexia, and pathological gambling. Over the years, participants have contributed importantly to the development of detection models and exploited the datasets we created to address these critical disorders. In 2025, for the ninth edition of the lab, we introduce new tasks designed to improve risk detection actual approaches through deeper contextual and conversational analysis. This year’s eRisk lab includes the continuation of the sentence ranking task for depressive symptoms, along with two new tasks: contextualized early detection of depression, which leverages full conversational interactions, and a pilot task on detecting depression using fine-tuned conversational agents. This paper outlines our progress to date, shares insights from previous editions, and details our innovative plans for eRisk 2025.