Artificial Intelligence in Predicting Therapeutic Responses to Biologics for Inflammatory Bowel Disease:a Review
Early identification of therapeutic responses to inflammatory bowel disease(IBD)is one of the challenges in clinical practice.This article reviews how artificial intelligence(AI)has been applied in predicting therapeutic responses to IBD biologics,covering the application of various machine learning and deep learning algorithms in efficacy prediction models for tumor necrosis factor(TNF)-α agents,ustekinumab(UST),and vedolizumab(VDZ).It has been found that clinical characteristics and laboratory tests are the most common predictive factors,and that compared to models that only include these two factors,models that incorporate endoscopic scores,multi-omics,and imaging data demonstrate better performance.Random forest(RF)is the most commonly used AI model,followed by artificial neural networks(ANN),with similar model performance for both.AI models have demonstrated good performance in predicting clinical symptoms,inflammatory markers,and endoscopic mucosal manifestations after treatment.However,the indicators for predicting outcomes are relatively limited,lacking a more systematic evaluation of treatment efficacy.Additionally,as treatment goals such as transmural healing and histological remission gradually enter the research spotlight,there is anticipation for AI to assist in exploring optimal treatment outcomes and uncovering more potential predictive factors for clinical application.
ulcerative colitisCrohn's diseaseartificial intelligenceanti-tumor necrosis factor-α therapyustekinumabvedolizumabtherapyprediction model
NETL
NSTL
万方数据
