Forecasting best performance with incomplete resources: automated risk assessment of school violence

School violence risk prevention relies on manual assessments that are time-consuming and subjective. We developed a risk assessment protocol involving two sets of interview questions: 1) Brief Rating of Aggression by Children and Adolescents (BRACHA); and 2) School Safety Scale (SSS). We developed a machine learning algorithm to predict risk levels using natural language processing (NLP) of interview transcripts. We evaluated the incremental change in performance with the consecutive addition of each question to simulate situations where interviews cannot be completed, and we evaluated the algorithm for bias.