Designing a System for Evaluating at Risk Individuals Celiac Disease

Abstract

Abstract Introduction: Celiac disease is an autoimmune gastrointestinal disease that causes disruption of nutrient absorption by damaging the villi of the small intestine. The progression of celiac disease in Iran is increasing, and the lack of timely detection causes various complications. Data mining and machine learning algorithms appear to be necessary to predict this disease. Our goal in this study is to develop a clinical decision-support tool to identify the risk of celiac disease in persons. Materials and methods : This research is a developmental applied study, researchers collected demographic information about all people referred to Celiac Centers of Shahid Beheshti in Tehran and Imam Reza in Tabriz between 2015 and 2018 by census method. Decision tree algorithm was selected and modeled based on the results of data processing. Finally, the analysis of the results was done using the confusion matrix and the area under the graph, and the applicability of the system was measured with Quiz questionnaire. Findings: According to the obtained results and the optimal experience of using random forest and its proper performance, the random forest model was used in this research. The detection ability of this algorithm based on the method of evaluating the area under the diagram, respectively, is 98% low risk, 99% average risk, and 99% high risk in people. This system was evaluated as good by users based on the results of the quiz questionnaire and received an overall rating of 87.8. Experts all agreed that this system was 100% effective for improving disease diagnosis. Conclusion: The proposed prediction model based on machine learning has the ability to identify the risk of celiac disease in people and can be implemented in the real environment, which is considered a clinical support tool needed by doctors. Keywords: Celiac disease, Decision support system, Risk assessment, Machine learning