The field of clinical research is witnessing a significant shift towards automation and computational phenotyping, driven by the adoption of large language models (LLMs). Recent developments have demonstrated the potential of LLMs in improving the efficiency and accuracy of clinical trial adjudication, computational phenotyping, and data extraction from unstructured clinical data.

The use of LLMs has shown promise in reducing the time and costs associated with manual data review, while maintaining high-quality and consistent outcomes. Furthermore, the development of novel evaluation frameworks and hybrid models has enabled the improvement of computational phenotyping processes and the identification of immune checkpoint inhibitor studies in genomic repositories.

Noteworthy papers include: Automating Adjudication of Cardiovascular Events Using Large Language Models, which presents a novel framework for automating the adjudication of cardiovascular events in clinical trials. PHEONA: An Evaluation Framework for Large Language Model-based Approaches to Computational Phenotyping, which outlines context-specific considerations for evaluating LLM-based approaches to computational phenotyping. ProtoBERT-LoRA: Parameter-Efficient Prototypical Finetuning for Immunotherapy Study Identification, which achieved an F1-score of 0.624 in identifying immune checkpoint inhibitor studies. Evaluating Large Language Models for Automated Clinical Abstraction in Pulmonary Embolism Registries, which demonstrated the strong potential of LLMs in automating PE registry abstraction. ELM: Ensemble of Language Models for Predicting Tumor Group from Pathology Reports, which introduced a novel ensemble-based approach leveraging both small and large language models to predict tumor groups from pathology reports.