New research reported that natural language processing (NLP) can capture clinically relevant information more accurately than ICD-10 coding, a finding aimed at improving downstream clinical analytics and decision support. The work highlights an informatics limitation: ICD-10 codes are broad and sometimes lag clinical nuance, while NLP can extract detail directly from unstructured notes and other text-based records. For biotech drug development teams, better phenotyping and extraction can improve cohort selection for trials, support real-world evidence analysis, and reduce label noise in model training. As healthcare data platforms modernize, the key test will be integration—whether NLP pipelines can be deployed reliably across institutions, with consistent governance and auditing.