Performance of models predicting.

Objective
The prevalence of arteriovenous fistula (AVF) dysfunction and its associated complications among patients undergoing maintenance hemodialysis (MHD) underscores the necessity for predictive models aimed at early diagnosis and intervention. Although various models for predicting AVF risk have emerged, a comprehensive review of their advancements and challenges is currently lacking. This study aims to systematically evaluate risk prediction models for AVF failure in MHD patients, thereby providing insights for clinical practitioners in selecting or developing more effective risk assessment tools.
Methods
A systematic search was conducted across multiple databases, including PubMed, Embase, Web of Science, Cochrane Library, ClinicalTrials.gov registry platform,the China Biomedical Literature Database, CNKI, WanFang Database, and the VIP Database, focusing on studies related to risk prediction models for AVF failure in MHD patients. Google Scholar was also searched to retrieve gray literature. The search encompassed literature from the inception of these databases up to August 1, 2024. Two independent researchers performed literature screening and data extraction, utilizing the Prediction Model Risk of Bias Assessment Tool to evaluate the methodological quality of the included studies.
Results
Initially, 2,052 studies were identified. After thorough screening, 11 studies were ultimately included, detailing 11 distinct risk prediction models for AVF failure in MHD patients. The sample sizes of these studies ranged from 126 to 14,892 participants. The identified models exhibited varying predictive performances, highlighting common limitations such as small sample sizes, improper handling of missing data, and a lack of external validation.
Conclusion
The risk prediction models for AVF failure among MHD patients demonstrate adequate predictive performance; however, the overall quality of the research necessitates improvement. Future studies should prioritize refining research design and reporting processes, as well as validating and enhancing existing models to ascertain their effectiveness and feasibility in clinical practice.