A call triage support system for emergency medical service using multiple random forests

This paper proposes a call triage system for emergency medical service with multiple random forests (RFs). A call triage is an operation to determine the level of first-aid service quality required based on the severity and urgency of a victim. The aim is to achieve higher overall performance and clarify the method for deciding the determination thresholds while simultaneously achieving both fail-safe determination and overall accuracy. To achieve the former aim, we apply the RF to determine the category. Furthermore, to achieve the latter aim, we propose a new algorithm for determining the severity using multiple RFs. The results of evaluation experiment show that the proposed algorithm can provide both accuracy and fail-safe operation in the determination of severity.