Evaluation of a Diagnostic Decision Support System for the Triage of Patients in a Hospital Emergency Department

Author
Keywords
Abstract
One of the biggest challenges for the management of the emergency department (ED) is to expedite the management of patients since their arrival for those with low priority pathologies selected by the classification systems, generating unnecessary saturation of the ED. Diagnostic decision support systems (DDSS) can be a powerful tool to guide diagnosis, facilitate correct classification and improve patient safety. Patients who attended the ED of a tertiary hospital with the preconditions of Manchester Triage system level of low priority (levels 3, 4 and 5), and with one of the five most frequent causes for consultation: dyspnea, chest pain, gastrointestinal bleeding, general discomfort and abdominal pain, were interviewed by an independent researcher with a DDSS, the Mediktor system. After the interview, we compare the Manchester triage and the final diagnoses made by the ED with the triage and diagnostic possibilities ordered by probability obtained by the Mediktor system, respectively. In a final sample of 214 patients, the urgency assignment made by both systems does not match exactly, which could indicate a different classification model, but there were no statistically significant differences between the assigned levels (S = 0.059, p = 0.442). The diagnostic accuracy between the final diagnosis and any of the first 10 Mediktor diagnoses was of 76.5%, for the first five diagnoses was 65.4%, for the first three diagnoses was 58%, and the exact match with the first diagnosis was 37.9%. The classification of Mediktor in this segment of patients shows that a higher level of severity corresponds to a greater number of hospital admissions, hospital readmissions and emergency screenings at 30 days, although without statistical significance. It is expected that this type of applications may be useful as a complement to the triage, to accelerate the diagnostic approach, to improve the request for appropriate complementary tests in a protocolized action model and to reduce waiting times in the ED.
Year of Publication
2019
Journal
International Journal of Interactive Multimedia and Artificial Intelligence
Volume
5
Issue
Special Issue on Artificial Intelligence Applications
Number
4
Number of Pages
60-67
Date Published
03/2019
ISSN Number
1989-1660
Citation Key
URL
DOI
Attachment