<p>Intradialytic hypotension (IDH) is one of the most common complications of hemodialysis. In order to prevent IDH, Artificial intelligence (AI) was utilized in predicting the onset of IDH. Different variables and different AI methods were used to improve the diagnostic test accuracy but resulted in different findings. We conducted a systematic review and meta-analysis to quantitatively analyze the diagnostic test accuracy of AI models in predicting IDH in HD patients and investigated the factors influencing the diagnostic test accuracy. Medline, Embase, and the Cochrane Library were searched using combined free texts and MeSH terms relating to hemodialysis, hypotension and AI (prior to Sep, 2025 and English-language publications included). Two authors independently screened the studies, and extracted and analyzed the data. A quantitative random-effects meta-analysis using Rutter and Gatsonis hierarchical summary receiver operating characteristics (HSROC) model was applied to summarize the SROC curves. Between-study heterogeneity was explored using multiple methods, and funnel plot analysis was used to identify publication bias. A total of 142 articles were identified and 19 studies with 33,214(95%CI: 405.08 to 3065.86) subjects and 4,350,177(95%CI: 0 to 35865.73) sessions met the inclusion criteria. Decision tree, support vector machine, XGBoost, extreme gradient boosting, recurrent neural network, and random forest were used in the studies. The summary area under the SROC curve(AUC) was 0.86(95%CI: 0.83 to 0.89), showing that the sensitivity of AI for prediction of IDH was 0.74(95%CI: 0.66 to 0.82), specificity was 0.85 (95%CI: 0.77to 0.91).These results indicate that it has moderate diagnostic ability for prediction of IDH during hemodialysis and high accuracy in excluding negative patients, although some missed diagnosis remain unavoidable. The Deeks’ funnel plot asymmetry test showed no evidence of publication bias (<i>P</i> = 0.68). This comprehensive systematic review and meta-analysis of AI models demonstrates promising accuracy for predicting IDH in hemodialysis patients. However, current limited studies and between-study heterogeneity limit its clinical application. More standardized approaches are needed for the development and validation of prediction models and randomized clinical trials, including unified IDH definition criteria, standardized data collection and model development protocols, and consistent external validation standards, to evaluate the clinical value of IDH prediction models.</p>