Overview
The association between rash and survival outcomes in patients treated with first- and second-generation epidermal growth factor receptor tyrosine kinase inhibitors (EGFR TKIs) is well-documented, but less is known about third-generation TKIs. This study leveraged machine learning (ML)-extracted real-world adverse events (rwAEs) to evaluate the incidence of rash and its relationship with survival outcomes in patients with advanced non–small cell lung cancer (NSCLC) treated with EGFR TKIs.
Using the Flatiron Health EHR-derived US database, researchers analyzed data from over 5,600 patients treated between 2011 and 2024 and utilized a natural language processing model to extract rwAEs. The study found that rash was associated with improved real-world overall survival (rwOS) and progression-free survival (rwPFS) across all TKI generations, including third-generation TKIs. Additionally, ML extraction combined with ICD codes identified a higher rash incidence compared to ICD codes alone, while survival benefits remained consistent across methods.
Why this matters
This study highlights the ability of ML to scalably extract rwAEs, potentially improving the accuracy and completeness of clinical data. The findings confirm that rash remains a positive prognostic indicator for survival in patients treated with EGFR TKIs, including third-generation therapies. By demonstrating the effectiveness of ML in identifying rwAEs, this research supports the broader adoption of AI-driven methodologies in oncology, ultimately enhancing real-world evidence generation.