Gaining insights into the utilization and effectiveness of oral cancer therapies within electronic health records is crucial for identifying unmet needs and advancing drug development. However, the manual abstraction of this information from unstructured data by clinical experts is a time-consuming and resource-intensive task.
In this study, researchers aimed to investigate the influence of two different approaches to data curation, namely expert-abstraction and machine learning (ML) based NLP (ML-extraction), on the ability to accurately measure patient characteristics and capture real-world treatment patterns.
Why this matters
This study aimed to demonstrate that employing ML-extracted variables on expert-abstracted oncology data can yield comparable results in downstream analysis to using manual abstracted variables directly. By achieving similar outcomes, this approach has the potential to enable a comprehensive understanding of drug utilization and patient treatment profiles on a larger scale.