Automated Extraction of Substance Use and Co-occurring Disorders from Probation Records

The authors describe the application of advanced Natural Language Processing (NLP) and Artificial Intelligence (AI) methods to automatically discover and analyze important information from text. The system had been applied to probation documents for over 98,000 individuals under supervision as of October 2021, analyzing over 62 million sentences to detect history of substance use disorder, diagnosis of co-occurring mental conditions, and participation in any past or ongoing treatment program. The system performs PDF content extraction, in-depth linguistic analysis, and temporal reasoning to discover the individuals’ life events and the entities, participants, and temporal relations associated with the events.