Identification of Gout Flares in Chief Complaint Text Using Natural Language Processing

Area of Science:

• Medical Informatics
• Clinical Research
• Rheumatology

Background:

• Patients with gout flares treated in the Emergency Department (ED) often lack continuity of care post-visit.
• Effective identification and referral to outpatient gout management are needed.
• Prospective identification of gout flares in the ED is challenging due to minimal documentation.

Purpose of the Study:

• To develop and evaluate a method for prospectively identifying gout flares in ED patients.
• To assess the predictive power of chief complaint notes for gout flares.
• To create and share a valuable clinical corpus for Natural Language Processing (NLP) research.

Main Methods:

• Annotation of ED triage nurse chief complaint notes for gout flares.
• Implementation of a simple algorithm for generating gout flare ED alerts.
• Creation of a de-identified corpus of free text chief complaint notes.

Main Results:

• Chief complaint notes alone demonstrate strong predictive power for identifying gout flares.
• A functional algorithm for ED gout flare alerts was developed.
• A novel, de-identified clinical corpus of chief complaints was made publicly available.

Conclusions:

• Chief complaints in ED triage notes are highly effective for prospective gout flare identification.
• This approach can facilitate timely referrals and improve continuity of care for gout patients.
• The released corpus supports further NLP research in clinical settings.