Predicting inflammatory location of Crohn's disease using stool microbiome profiles

Abstract

Crohn’s disease is a crucial disease which is increasing in developed country. As Crohn’s disease is a complex disease, the diagnosis process is also complex and time-consuming. To solve the problem, I suggest a new method to find the location of Crohn’s disease using stool sample. To reduce the complexity of microbiome data, scaling, outlier detection, and feature selection was conducted. By using embedding method, microbiome data which is compressed to be suitable for classification problem. The prediction score was highest in my method and based on the classifier feature importance score, I suggested a new microbe which can be important in Crohn’s disease diagnosis and treatment.