Research described in a new report says Natural Language Processing (NLP) tools outperform ICD-10 coding for capturing clinically relevant information, a finding aimed at improving the fidelity of downstream real-world evidence. The story frames the comparison around accuracy in capturing clinical data from records, suggesting NLP approaches can better extract meaning than the long-standard diagnosis coding system. For pharma and healthcare analytics teams, this could affect how endpoints and comorbidity signals are derived when ICD-10 is used as a proxy for clinical features. If the performance gap holds across settings, sponsors may need to revisit data pipelines that rely on code-based abstraction and consider NLP-driven extraction to reduce missingness or misclassification. The report emphasizes informatics implications rather than new therapeutics.