The causes of abdominal pain can be numerous and vary according to a particular quadrant. As far as the anatomy of the upper abdomen is concerned, the organs mainly involved are the stomach, spleen, liver, pancreas, gallbladder, and bile ducts. Causes of pain can be numerous, including gall bladder torsion, choledocholithiasis, cholelithiasis, pancreatitis, cholangitis, cholecystoenteric fistula, liver abscess, hepatocellular carcinoma, gastritis, etc. Cholecystitis is the main source of upper right quadrant pain and is the typical reason for patient admission with acute abdominal pain. Regarding upper abdominal pain, the cause can be related to either stomach, spleen, liver, pancreas, gallbladder, or bile ducts. For stomach-related disorders, the preferred modality is CT. For spleen or liver-related disorders, the preferred modality can be USG. The preferred modality for the pancreas, gallbladder, and bile duct is MRI. Thus, it is clear that the choice of modality on the basis of suspected cause varies and is more challenging. In such cases, there might be chances of human errors in deciding the choice of modality. The aim of the work is to overcome such human errors via machine learning. Data mining has become an ease of use for predicting modality from recent years as far as healthcare sectors are concerned. Data mining acquires the quality of dredging details from large amounts of sets of data, warehouses, or other depositories. Predicting modality on the basis of symptoms from the capacious medical databank is challenging for researchers up to a great extent. Thus, in order to sort this matter, the researchers make use of data mining techniques, which include classification, clustering, association rules, etc. The major purpose of the given research work is to forecast the particular modality on the basis of symptoms using classification algorithms. Naive Bayes algorithm is used in this work. The result of this study was obtained by Naive Bayes probabilistic classifier with an expert diagnostic result. Modality percentage accuracy based on the experimental results has been acquired by Naive Bayes-based expert system. So, the conclusion is that an expert system in combination with Naive Bayes possesses the great capacity of using successfully by people however improvement is still needed to attain further success rate.
