The difficult diagnosis of acute appendicitis of patients appealing to the hospital with abdominal pain often leads to unnecessary acute appendicitis operations. Accordingly, the aim of this study is to be able to provide the correct diagnosis whether the existing case indeed necessitates operation or not through machine learning algorithms based on classification. To that purpose, SMOTE, random oversampling, and random undersampling methods were proposed to reduce the negative effects of imbalanced data set problem on classification, and it was benefitted from the risk factors in relation to Alvarado Score to predict the diagnosis of acute appendicitis. Additionally, a classification model was generated by using support vector machine classification algorithm. A decision support system was developed that could contribute to the decision making by generating interface for support vector machine algorithm in which the best performance was obtained.
TopIntroduction
Acute appendicitis (AA) is the most common emergency in patients appealing to the hospital with abdominal pain in the healthcare field, and appendectomy (acute appendicitis surgery) is one of the first emergency operations worldwide (Andersson, Jaffe, & Berger, 2014). For this reason, the decision of whether the patient should be operated or not should be taken quickly. It is difficult to diagnose acute appendicitis. Negative appendectomy, which is determined as a result of the examination of the postoperative material, is seen to cause dozens of unnecessary operations and patients suffering due to the rate of encountering a disease-free appendix during surgery. A false negative diagnosis can delay the diagnosis and therefore increase morbidity, mortality and legal problems. (Flum & Koepsell, 2008). However, clinical diagnosis is often a challenge even for experienced surgeons. Negative appendectomy rate reaching 20-30% proves how difficult it is. (Yang, Wang, Chung, & Chen, 2006). It has been reported that the clinical diagnosis of acute appendicitis was correctly diagnosed in 78% of men and 58% in women. (Pieper, Keger, & Niisman, 1982). Many patients who underwent appendectomy were found to have negative histopathology of the surgically removed appendix, which is the gold standard for the diagnosis of appendicitis (Munir, Mushtaq, Ishaque, Mudassar, & Khalid, 2008). In the literature, it has been reported that the negative laparotomy rate for AA ranges between 4-75% (Eskelinen & Lipponen, 1992). In recent years, the decision of whether or not patients to undergo surgery poses a challenge with negative appendectomy rates of 10-25%, causing dozens of unnecessary operations (Hale, Moolloy, Pearl, Schutt, & Jaques, 1997). On the other hand, available statistics show that 1 out of 5 acute appendicitis cases were misdiagnosed and 40% of patients who underwent emergency appendectomy had a normal appendix (Kwok, Kim, & Gorelick, 2004). In another study, the possibility of encountering a negative laparotomy was reported to be between 13 and 36% (Bergeron, 2006). In another study, negative appendectomy rates of 3-30% of perforated appendicitis rates is seen (Binnebösel, Otto, Manhken, Gassler, & Schumpelick, 2009). It is seen in a different study that the negative appendectomy rates varied between 14.7% and 8.47% despite the advanced laboratory tests and imaging techniques in today's conditions (Seetahal, Bolorunduro, & Sookdeo, 2011). In another study, it is stated that the negative appendectomy rates determined as a result of the examination of the postoperative material are generally between 3-4% and 44% (Cleeve, Jones, Joshi, & Ward, Trends in Childhood appendicitis, 2011). Early surgical interventions without definitive diagnosis can result in a negative appendectomy rate of 20-30%, especially in women of childbearing age (DeKoning, 2016).
The removal of a normal appendix is a burden for both patients and healthcare resources (Khan, 2006). The increase in negative appendectomy rates can cause unnecessary morbidity and complications, and with the increase in treatment costs, physicians may face lawsuits (Liang, Andersson, & Jaffe, 2014).
Acute appendicitis appears to mimic many diseases such as acute abdominal pain, intestinal obstructions, gastroenteritis, acute cholecystitis, ureter stone colic, pancreatitis, diverticulitis, urinary tract infections, cystitis, lymphadenopathies, intestinal ischemia (Humes & Simpson, 2006).
This study emerged as a result of the mentioned reasons, negative appendectomy rates and interviews with specialist physicians.
The data sets of the patients in this study, which was carried out to reduce unnecessary AA operations due to misdiagnosis, bring about the class imbalance problem, which is frequently encountered in the field of health.