[Application of artificial intelligence systems in the emergency room : Do the communication patterns give indications for possible starting points? An observational study].

Background: High expectations are currently attached to the application of artificial intelligence (AI) in the resuscitation room treatment of trauma patients with respect to the development of decision support systems. No data are available regarding possible starting points for AI-controlled interventions in resuscitation room treatment.

Objective: Do information request behavior and quality of communication indicate possible starting points for AI applications in the emergency room?

Material And Methods: A 2‑stage qualitative observational study: 1. Development of an observation sheet based on expert interviews that depicts the following six relevant topics: situational factors (course of accident, environment), vital parameters, treatment-specific Information (treatment carried out). trauma-specific factors (injury patterns), medication, special features of the patient (anamnesis, etc.) 2. Observational study Which topics were inquired about during emergency room treatment? Was the exchange of information complete?

Results: There were 40 consecutive observations in the emergency room. A total of 130 questions: 57/130 inquiries about medication/treatment-specific Information and vital parameters, 19/28 of which were inquiries about medication. Questions about injury-related parameters 31/130 with 18/31 regarding injury patterns, course of accident (8/31) and type of accident (5/31). Questions about medical or demographic background 42/130. Within this group, pre-existing illnesses (14/42) and demographic background (10/42) were the most frequently asked questions. Incomplete exchange of information was found in all six subject areas.

Conclusion: Questioning behavior and incomplete communication indicate a cognitive overload. Assistance systems that prevent cognitive overload can maintain decision-making abilities and communication skills. Which AI methods can be used requires further research.