Natural language processing identifies symptoms predicting complete coronary artery occlusion in patients with NSTEMI

Area of Science:

• Cardiology
• Medical Informatics

Background:

• Acute coronary syndrome (ACS) affects 10% of emergency department visits.
• The 13-item ACS Symptom Checklist aids rapid symptom assessment.
• Distinguishing between occluded and non-occluded arteries in NSTEMI is crucial.

Purpose of the Study:

• To evaluate the 13-item ACS Symptom Checklist's effectiveness in differentiating NSTEMI patients with and without occluded arteries.
• To utilize natural language processing (NLP) for symptom extraction and analysis.

Main Methods:

• Retrospective analysis of 1905 NSTEMI patients.
• Extraction of 13 ACS symptoms using NLP.
• Chi-square tests and logistic regression (sex-stratified) to assess symptom predictive value.
• Outcome defined as coronary artery requiring revascularization.

Main Results:

• Arm pain, unusual fatigue, and palpitations differentiated patients with and without occluded arteries.
• Arm pain showed a significant association (OR 1.532) with occluded arteries, with similar odds in men and women.
• NLP successfully identified key symptoms from clinical notes.

Conclusions:

• Arm pain is a primary symptom predicting occluded coronary arteries in NSTEMI patients.
• Minimal sex differences were observed in the predictive value of arm pain.
• NLP is a valuable tool for symptom analysis in clinical notes.