Incorporating unstructured data sources & natural language processing (NLP) into medical research provides a more accurate view of patient health. 

The study “Assessing Prevalence of Hypoglycemia in a Medical Transcription Database” utilized the AnswerY™ unstructured database and revealed significant underreporting of hypoglycemia in traditional medical data sources, like claims and electronic health record (EHR) databases. It found instances of hypoglycemia to be 2 to 9 times higher than previously reported.

Why it matters: This discrepancy highlights the limitations of structured claims-based + EMR data in capturing the full scope of patient health concerns, particularly those with diabetes. 

The big picture: The study’s use of NLP to analyze medical transcripts suggests that unstructured data, when mined effectively, can offer a more accurate view of patient health issues like hypoglycemia. 

By the numbers: Prevalence of hypoglycemia was estimated at 18% among patients with type 1 diabetes and 8% among patients with type 2 diabetes, significantly higher than the 1% to 4% suggested by traditional data sources like claims or EHR databases. 

The bottom line: The findings underscore the value of incorporating unstructured data sources + advanced analytics into medical research to ensure a comprehensive understanding of patient health. 

Zoom in: AnswerY, our proprietary qualitative database and platform, houses over 80 million HIPAA-compliant, unstructured patient-provider interactions and insights from 150,000+ healthcare providers. This rich dataset uncovers the “why” behind treatment decisions, including reasons for starting, switching, or stopping treatment. By leveraging advanced NLP techniques, AnswerY extracts valuable information from unstructured medical records, providing a more holistic view of patient care.