Analyzing Patient Secure Messages Using a Fast Health Care Interoperability Resources (FIHR)-Based Data Model:...

Area of Science:

• Health Informatics
• Natural Language Processing
• Medical Data Modeling

Background:

• Patient portals linked to EHRs have grown since the Medicare Access and Children's Health Insurance Program Reauthorization Act and Meaningful Use program.
• Secure messaging via patient portals has increased, offering research opportunities in patient-centered care.
• Analyzing patient secure messages is crucial for understanding patient needs and improving healthcare delivery.

Purpose of the Study:

• To develop a data model for patient secure messages based on the Fast Healthcare Interoperability Resources (FHIR) standard.
• To identify and extract significant information from patient secure messages.
• To facilitate natural language processing (NLP) analysis of patient communications.

Main Methods:

• Iterative development and annotation of a data model using FHIR standards.
• Manual review and annotation of over 2 million patient secure messages.
• Application of topic modeling to identify hidden themes in patient messages.

Main Results:

• A 3-level hierarchical data model with 3 macroconcepts, 28 mesoconcepts, and 85 microconcepts was created.
• Clinical concepts formed the largest category (64.38%) in the annotated corpus.
• Topic modeling successfully identified key concepts like fatigue, prednisone, and patient visits, with 89.2% of keywords aligning with the data model.

Conclusions:

• The developed data model and annotated corpus enable better identification and understanding of medical concepts in patient messages.
• The model can be extended to analyze other patient narratives from social media and forums.
• Future work includes developing NLP solutions for automated triaging and enhanced patient-centered care analysis.