Introduction Acute abdomen is an abnormal condition characterized by sudden onset of severe pain within
the abdomen, which requires immediate diagnosis and treatment. From different acute abdominal
diseases, intestinal obstruction accounts for a great proportion of morbidity and mortality in Africa including Ethiopia. This is due to late presentations, improper health care facilities, and the cause of abdominal pain is not identified timely.
Objective This study aimed to develop a knowledge-based system for the diagnosis and treatment recommendation of selected acute abdominal diseases by collaborating data mining results with the knowledge obtained from the domain experts.
Methodology Design science research methodology (DSRM) used to conduct this research. To extract the required knowledge, the researchers have collected 2260 instances of intestinal obstruction dataset from Debre Berhan referral hospital, and used semi-structured interview technique to acquire from the domain experts. To develop the classifier model, the researchers have conducted four experiments with four different scenarios by using Naïve Bayes, J48, JRip, and PART algorithms.
Result The best performance result achieved by applying PART algorithm on selected attributes under the 10-fold cross-validation test option with 98.19% accuracy. To acquire knowledge from the experts the researchers have used semi-structured interview and used purposive sampling to select domain experts. Finally, After the required knowledge was discovered, the knowledge-based system was developed and the developed system was evaluated Lastly, the evaluation result shows that 91.3% accuracy for the system performance by using test cases and 90.6% for the user acceptance testing.
Conclusion applying the developed knowledge-based system will reduce the morbidity and mortality rate caused by intestinal obstruction. And it will give a significant advantage to the patients, doctors and for the healthcare organizations.