A case study of forensic psychiatry experts' reports analysis through large language models

Abstract

The integration of artificial intelligence (AI) technologies in forensic psychiatry has gained significant attention due to their potential to enhance tasks such as outcome prediction and decision-making. In this study, we explored the feasibility and performance of a large language model (LLM-GPT) in extracting both clinical and non-clinical variables from authentic forensic psychiatric reports concerning defendants' criminal responsibility and social dangerousness. We employed GPT-4o to extract relevant data using a set of custom queries, which we applied to two forensic psychiatric expert reports. The results of the study demonstrated that the system was capable of extracting information from the forensic psychiatric reports and generating a summarized version. Identifying the most important parts to construct a meaningful synthesis in a highly specialized application domain is currently a challenge. This study highlights the potential of AI in forensic psychiatry and suggests that this approach could be valuable for collecting semi-automated or automated data from reports, enabling the creation of a large dataset that could be used for further research and analysis.