P57 Machine Learning-Accelerated Outcomes Research: A Real-World Case Study of Biomarker-Associated Overall Survival in Oncology

Meaningful outcomes research requires unstructured data found in electronic health records (EHRs) which are often missing from administrative claims. Historically, clinical experts manually review charts, a resource intensive process. Researchers have developed machine learning (ML) models to recognize patterns in language documenting characteristics of interest and extract clinically relevant information. We explored the impact of data curation method (expert-abstraction vs ML-extraction) on the association between real-world overall survival (rwOS) and ROS1 rearrangement status in advanced non-small cell lung cancer (aNSCLC).