Research2h ago
LLMs Extract Cancer Prognostics From Clinical Notes
Researchers analyzed records from 3,500 lung and colon cancer patients to extract prognostic data from free-text notes. Using LLMs without task-specific training, the team identified mobility impairment and comorbidities missed by structured EHRs. This approach allows clinicians to capture nuanced patient characteristics. It proves that general-purpose models can outperform traditional scoring systems in oncology.