TY - GEN
T1 - Application of an artificial intelligence method for diagnosing acute appendicitis
T2 - 8th FTRA International Conference on Future Information Technology, FutureTech 2013
AU - Park, Sung Yun
AU - Seo, Jun Seok
AU - Lee, Seung Chul
AU - Kim, Sung Min
PY - 2014
Y1 - 2014
N2 - The aim of this study is to suggest an artificial intelligence model to diagnosis acute appendicitis using a support vector machine (SVM). Acute appendicitis is one of the most common abdominal surgery emergencies. Various methods have been developed to diagnose appendicitis, but they have not performed well in the Middle East, Asia, or the West. A total of 760 patients were used to construct the SVM. Both the Alvarado clinical scoring system (ACSS) and multilayer neural networks (MLNN) were used to compare performance. The accuracies of the ACSS, MLNN, and SVM were 54.87%, 92.89, and 99.61%, respectively. The areas under the curve of ACSS, MLNN, and SVM were 0.621, 0.969, and 0.997 respectively. The performance of the AI model was significantly better than that of the ACSS (P < 0.001). We consider that the developed models are a useful method to reduce both negative appendectomies and delayed diagnoses, particularly for junior clinical surgeons.
AB - The aim of this study is to suggest an artificial intelligence model to diagnosis acute appendicitis using a support vector machine (SVM). Acute appendicitis is one of the most common abdominal surgery emergencies. Various methods have been developed to diagnose appendicitis, but they have not performed well in the Middle East, Asia, or the West. A total of 760 patients were used to construct the SVM. Both the Alvarado clinical scoring system (ACSS) and multilayer neural networks (MLNN) were used to compare performance. The accuracies of the ACSS, MLNN, and SVM were 54.87%, 92.89, and 99.61%, respectively. The areas under the curve of ACSS, MLNN, and SVM were 0.621, 0.969, and 0.997 respectively. The performance of the AI model was significantly better than that of the ACSS (P < 0.001). We consider that the developed models are a useful method to reduce both negative appendectomies and delayed diagnoses, particularly for junior clinical surgeons.
KW - a receiver operating characteristics graph
KW - appendicitis
KW - artificial intelligence
KW - clinical scoring system
KW - support vector machine
UR - http://www.scopus.com/inward/record.url?scp=84899808436&partnerID=8YFLogxK
U2 - 10.1007/978-3-642-40861-8_13
DO - 10.1007/978-3-642-40861-8_13
M3 - Conference contribution
AN - SCOPUS:84899808436
SN - 9783642408601
T3 - Lecture Notes in Electrical Engineering
SP - 85
EP - 92
BT - Future Information Technology, FutureTech 2013
PB - Springer Verlag
Y2 - 4 September 2013 through 6 September 2013
ER -