Diagnosing acute appendicitis disease using support vector machine

Background and aims: One of the highest diagnostic possibilities for patients who come toemergency departments with abdominal pain is acute appendicitis, followed by one of the mostcommon emergency surgeries in general surgery, appendectomy. Despite the invention of variousdiagnostic methods, the rate of unnecessary appendectomy is significant. The use of artificial intelligenceand machine learning methods can improve the process of diagnosis and treatment. Inthe present study, the support vector machine system was used to help diagnose acute appendicitiswith the aim of increasing diagnostic accuracy and reducing the amount of unnecessary appendectomyand surgical outcomes.Method: During the research, by studying specialized texts on gastrointestinal diseases, effectivediagnostic variables were collected and categorized in the form of a checklist and evaluated andscored by experts. The research database includes ۱۴۲ cases of patients who underwent appendectomyin Taleghani Hospital in Tabriz during ۲۰۱۹. Then the support vector machine system withdifferent architectures was implemented and compared to determine the best diagnostic performance.The sensitivity, accuracy and specificity indices were used for evaluation.Results: The output obtained from the vector machine system for diagnosing acute appendicitishad sensitivity, accuracy and specificity of ۹۲.۷۹%, ۹۵.۴۳% and ۹۶.۵۸%, respectively.Conclusion: Considering the fact that the accuracy rate of diagnosis before surgery should beabove ۸۵%, the performance of the support vector machine system designed to diagnose acuteappendicitis is favorable and can help doctors in diagnosing acute appendicitis disease faster andmore accurately. Paying attention to the complications of late diagnosis of the disease, unnecessaryappendectomy, the duration of the patient’s stay in the hospital and its costs.

کلیدواژه ها:

Appendicitis ، diagnosis ، support vector machine ، artificial intelligence ، machine learning