Precision psychological markers enable targeted treatment in digital eating disorder interventions.

Digital mental health interventions for eating disorders can expand treatment access, but high attrition rates and heterogenous patient responses indicate a need for precision medicine approaches that identify which patients will benefit most and predict their treatment outcomes. We analyzed 30 days of Recovery Record (a widely-adopted digital intervention for eating disorders) data from 1,166 participants with lifetime bulimia nervosa or binge-eating disorder, with assessments at baseline, midpoint, and endpoint. We identified three symptoms (eating preoccupation, shape/weight preoccupation, and fear of losing control over eating) as precision psychological markers that predict and inform treatment response. These symptoms demonstrated the strongest correlations with overall symptom improvement measured by the Eating Disorder Examination Questionnaire (EDE-Q) Global score (r=0.65-0.66), predicted outcomes when elevated at baseline (all P<0.001), mediated treatment effects through early symptom changes (all P<0.001), and differentiated response groups in cluster analysis. Participants demonstrated significant improvements across all eating disorder domains at the population level (Global score Cohen's d=-0.80), but cluster analysis revealed three distinct response patterns: strong responders (35.2%) achieved mean Global score reductions of -1.50 points, moderate responders (46.4%) achieved -0.36 points, and non-responders (18.4%) showed minimal change (-0.06 points). This framework may enable identification of individuals most likely to benefit from a digital intervention and potentially supports early treatment monitoring, paralleling precision medicine advances where psychological marker-driven patient selection could improve treatment decisions. This mechanistic approach provides a generalizable framework for developing targeted digital mental health interventions across psychiatric disorders.