Digital traces of childhood maltreatment: A data-donation approach studying adolescent TikTok behaviours

Abstract While the link between social media use and adolescent mental health remains debated, evidence points to disproportionate risks for adolescents exposed to child maltreatment (CM). Existing research relies largely on self-report and cross-sectional designs, rarely examining platform-specific behaviour. Using a participatory digital data-donation approach, we assembled a corpus of TikTok archives of 129 adolescents (ages 13–18; 31.8% CM-exposed) linked to validated CM and depression measures. From 24 candidate indicators, LASSO stability selection identified ten consensus features. Under nested cross-validation, logistic regression classified CM status revealing fewer single contacts, higher follower counts, and more frequent searches as the strongest discriminating features. Longitudinally, higher follower counts and less posting activity were associated with elevated depressive symptoms six months later. These exploratory findings show digital data donation is feasible and can be linked to clinical outcomes in adolescents, offering a novel resource for studying digital behaviour.