Abstract
PURPOSE: Inflammatory bowel disease (IBD) is difficult to diagnose and classify. The purpose of this study is to establish an artificial intelligence model based on fecal multi-omics data for multi-classification diagnosis of IBD and its subtypes. MATERIALS AND METHODS: A total of 299 clinical cohort studies were included in this study, including 86 healthy people, 140 CD patients and 73 UC patients. Based on the idea of hierarchical modeling for different groups, we model the total population and the groups with self-evaluation of "very well" and "slightly below par", respectively. The original total features were fecal multi-omics data, including metagenomics, metatranscriptomics, proteomics, metabolomics, viromics, faecal calprotectin. The importance, collinearity and other feature engineering methods were used to evaluate the features. Finally, three individualized diagnosis models with less features and high accuracy were obtained. RESULTS: First, we screened 111 features to form the optimal feature set for the total population and established a three-classification individual diagnosis model with AUC of 0.83, which can simultaneously diagnose health, CD and UC. Secondly, according to the hierarchical modeling of the total population, we established two models for population with different self-evaluation. For "very well" population, we screened 59 features and established a three-classification diagnostic model with AUC of 0.85. For the self-evaluation population with "slightly below par", we finally included 22 features and established a three-classification diagnostic model with AUC of 0.84. Only metabolomics and metatranscriptomics features were included in the optimal feature sets. CONCLUSION: This study provides a valuable method for high accuracy, noninvasive diagnosis and subtype identification of IBD patients. Researchers can choose biomarkers in different models according to different self-evaluation of patients. Simple noninvasive fecal sampling can be used to detect metabolomics and metatranscriptomics data, thus replacing the tedious and painful clinical colonoscopy and biopsy procedures.