Artificial intelligence in emergency medicine: A scoping review

Area of Science:

• Medical Informatics
• Artificial Intelligence in Healthcare
• Emergency Medicine Research

Background:

• Growing investment in AI for medicine, but applications in emergency settings are unclear.
• This scoping review identifies and synthesizes literature on AI applications in emergency medicine.

Purpose of the Study:

• To conduct a scoping review of artificial intelligence applications in emergency medicine.
• To identify trends, interventions, and outcomes of AI in emergency care settings.

Main Methods:

• Scoping review following PRISMA guidelines.
• Searched Medline-OVID, EMBASE, CINAHL, and IEEE databases until February 28, 2020.
• Included studies self-classifying as AI interventions relevant to the emergency department (ED) with reported outcomes.

Main Results:

• 150 studies were included, mostly retrospective (82.7%).
• 37% of interventions focused on diagnosis, 12.7% on diagnostic imaging.
• AI interventions outperformed human clinicians in 8% of studies with comparators.

Conclusions:

• AI research in emergency medicine is increasing.
• Promising AI interventions exist for acute radiographic imaging and prediction-based diagnoses.
• Higher quality evidence is required to assess clinical outcomes of AI in emergency care.