TABLE 1.

Summary of studies using artificial intelligence in IBD etiology.

References	Published year	Aim of study	Type of AI	Number of subjects	Input variables (number/type)	Outcomes
Isakov et al., 2017	2017	Prioritization of IBD risk genes to detect candidate novel IBD-associated genes	Four different machine-learning classification models: rf, svmPoly, xgbTree, and glmnet	180 CD, 149 UC, 94 colorectal neoplasms, and 90 normal tissue	309/expression data from both array and RNA-seq data sets, GO, KEGG, and the Pathway Interactions Database terms	67 novel candidate IBD-risk genes
Yuan et al., 2017	2017	Screening for differential expressing genes among different clusters based on an IBD database to identify genes related to IBD	SMO	59 CD, 26 UC, and 42 normal samples	12,754/expression levels of 12,754 genes	21 candidate genes related to IBD
Menti et al., 2016	2016	Assessment of the predictive power of three BMLTs as classifiers for EIM in CD patients	3 BMLTs: NB, BART, and BN	152 patients with CD	12/disease characteristics, risk factors, and genetic variables	Accuracy: 89% (BN achieved the best performance)
Bottigliengo et al., 2019	2019	Determination of whether BMLT could improve EIM prediction	3 BMLTs: NB, BART, and BN	152 patients with CD	12/disease characteristics, risk factors, and genetic variables	Sensitivity: 66.0%, specificity: 69.0% (BART achieved the best performance)
Peng et al., 2015	2015	Prediction of IBD onset and relapse frequency with meteorological data	ANN	569 UC and 332 CD patients	5/meteorological data	Accuracy in predicting the frequency of IBD relapse (mean square error = 0.009, mean absolute percentage error = 17.1%)

IBD, inflammatory bowel disease; CD, Crohn’s disease; UC, ulcerative colitis; GO, gene ontology; KEGG, Kyoto encyclopedia of genes and genomes; rf, random forest; svmPoly, support vector machine with polynomial kernel; xgbTree, extreme gradient boosting; glmnet, elastic net regularized generalized linear model; SMO, sequential minimal optimization; BMLTs, Bayesian Machine Learning Techniques; EIM, extra-intestinal manifestations; NB, Naïve Bayes; BART: Bayesian Additive Regression Trees; BN, Bayesian Networks; ANN, artificial neural network.