Automated identification of fall-related injuries in unstructured clinical notes - Takeaways - MDSpire

Automated identification of fall-related injuries in unstructured clinical notes

  • By

  • Wendong Ge

  • Lilian M Godeiro Coelho

  • Maria A Donahue

  • Hunter J Rice

  • Deborah Blacker

  • John Hsu

  • Joseph P Newhouse

  • Sonia Hernández-Díaz

  • Sebastien Haneuse

  • Brandon Westover

  • Lidia M V R Moura

  • July 26, 2024

  • 0 min

Share

  • 1

    Fall-related injuries (FRIs) significantly contribute to hospitalizations among older adults, costing over $50 billion in US healthcare spending in 2015.

  • 2

    Natural language processing (NLP) models were developed to automate the detection of FRIs in unstructured clinical notes from the Mass General Brigham health-care system.

  • 3

    Five NLP models, including BERT and RoBERTa, were trained and evaluated, with RoBERTa achieving the highest performance metrics for FRI identification.

  • 4

    The study utilized clinical notes from 2100 older adults, identifying 154,949 paragraphs of interest through automatic keyword scanning.

  • 5

    NLP models demonstrated high precision (0.90) and recall (0.91), indicating their potential to enhance research efficiency in identifying FRIs.

Original Source(s)

Automated identification of fall-related injuries in unstructured clinical notes

  • by Wendong Ge, Lilian M Godeiro Coelho, Maria A Donahue, Hunter J Rice, Deborah Blacker, John Hsu, Joseph P Newhouse, Sonia Hernández-Díaz, Sebastien Haneuse, Brandon Westover, Lidia M V R Moura

  • July 26, 2024

Related Content

Who Decides When a Doctor Steps Back?

Cognitive decline, credentialing, and why “I’ll know” often isn’t enough.

Evaluation of Physical Activity Levels and Sleep Issues in Older Adults with Mild Cognitive Impairment Using Smart Ring Technology

FDA Monthly Preview: Key April Decisions to Watch

Several high-stakes FDA decisions across multiple specialties are expected this month.