Yusuf Alper Kilic1,2, Ali Konan3, Kaya Yorganci3, Iskender Sayek3. 1. Department of General Surgery, Faculty of Medicine, Hacettepe University, 06100, Ankara, Turkey. yusufa@hacettepe.edu.tr. 2. Bilgitay Study Group, 06600, Ankara, Turkey. yusufa@hacettepe.edu.tr. 3. Department of General Surgery, Faculty of Medicine, Hacettepe University, 06100, Ankara, Turkey.
Abstract
INTRODUCTION: Trauma scoring aims for quantification and uniform reporting of trauma-related outcomes. Despite significant advances in trauma scoring, the exact time period at which relevant calculations should be made is not clear. Considering the importance of response to resuscitation, calculation of trauma scores after a period of resuscitation can allow better discrimination of patients who will survive. METHODS: A fuzzy-logic inference system, which is completely based on expert opinion and uses Glasgow Coma Scale (GCS) and systolic blood pressure at arrival to emergency room (ER) and their response to resuscitation as inputs, was developed. Records of the last 150 trauma patients admitted to our surgical intensive care unit (ICU) were used for calculations related to Injury Severity Score (ISS), Revised Trauma Score (RTS), Trauma and Injury Severity Score, and A Severity Characterization of Trauma (ASCOT) systems. Calculation of trauma severity and predicted mortality was performed at different time intervals during resuscitation [at arrival to emergency room (ER), after 1 h of resuscitation, and at ICU admission]. The performance of conventional systems and fuzzy-logic system was compared. RESULTS: Mean ISS was 32.31 ± 14.01. All systems included showed acceptable discriminative power. Among the conventional systems calculated at emergency room admission, ISS was the best performing [receiver operating characteristics (ROC), 0.9033] and RTS was the worst (ROC, 0.8106). Their performances were improved by up to 13% by use of post-resuscitation physiologic variables. Fuzzy-logic inference system performed slightly better (ROC, 0.9247) then the conventional systems calculated at arrival to ER. CONCLUSIONS: Response to resuscitation has significant impact on trauma mortality and must be considered in trauma scoring and mortality prediction. Fuzzy logic provides important opportunities for design of better predictive systems.
INTRODUCTION:Trauma scoring aims for quantification and uniform reporting of trauma-related outcomes. Despite significant advances in trauma scoring, the exact time period at which relevant calculations should be made is not clear. Considering the importance of response to resuscitation, calculation of trauma scores after a period of resuscitation can allow better discrimination of patients who will survive. METHODS: A fuzzy-logic inference system, which is completely based on expert opinion and uses Glasgow Coma Scale (GCS) and systolic blood pressure at arrival to emergency room (ER) and their response to resuscitation as inputs, was developed. Records of the last 150 traumapatients admitted to our surgical intensive care unit (ICU) were used for calculations related to Injury Severity Score (ISS), Revised Trauma Score (RTS), Trauma and Injury Severity Score, and A Severity Characterization of Trauma (ASCOT) systems. Calculation of trauma severity and predicted mortality was performed at different time intervals during resuscitation [at arrival to emergency room (ER), after 1 h of resuscitation, and at ICU admission]. The performance of conventional systems and fuzzy-logic system was compared. RESULTS: Mean ISS was 32.31 ± 14.01. All systems included showed acceptable discriminative power. Among the conventional systems calculated at emergency room admission, ISS was the best performing [receiver operating characteristics (ROC), 0.9033] and RTS was the worst (ROC, 0.8106). Their performances were improved by up to 13% by use of post-resuscitation physiologic variables. Fuzzy-logic inference system performed slightly better (ROC, 0.9247) then the conventional systems calculated at arrival to ER. CONCLUSIONS: Response to resuscitation has significant impact on trauma mortality and must be considered in trauma scoring and mortality prediction. Fuzzy logic provides important opportunities for design of better predictive systems.
Authors: C Healey; Turner M Osler; Frederick B Rogers; Mark A Healey; Laurent G Glance; Patrick D Kilgo; Steven R Shackford; J Wayne Meredith Journal: J Trauma Date: 2003-04