Using AI to Improve Operations in the Emergency Department

Abstract

Overcrowding in emergency departments (EDs) is a major problem in most hospitals, and long waiting times are one of its main causes. This problem is due to the complete reliance on manual processes, from registering the patient in the system to the triage and prioritization of examination. Thus, this paper proposes an electronic system as an alternative to manual procedures to mitigate the effects of overcrowding. The proposed system automates patient registration using an identity verification scanner and introduces an AI-based triage concept to support patient classification based on vital data entered by nurses. This approach assists in the classification of cases as emergency or non-emergency, and directly determines the course of action, either by immediately referring the patient to a doctor or waiting for the patient’s turn to be examined. A business process analysis (BPA) methodology was used to document and analyze both the current (“as-is”) and proposed (“to-be”) processes in the ED. A simulation was performed using the Bizagi Modeler to evaluate the performance of the proposed system. The results show that the redesigned process reduces the average patient waiting time in the ED by improving case periodization and workflow efficiency.