AimThis study aimed to establish a prediction model in peritoneal dialysis patients to estimate the risk of technique failure and guide clinical practice.MethodsClinical and laboratory data of 424 adult peritoneal dialysis patients were retrospectively collected. The risk prediction models were built using univariate Cox regression, best subsets approach and LASSO Cox regression. Final nomogram was constructed based on the best model selected by the area under the curve.ResultsAfter comparing three models, the nomogram was built using the LASSO Cox regression model. This model included variables consisting of hypertension and peritonitis, serum creatinine, low-density lipoprotein, fibrinogen and thrombin time, and low red blood cell count, serum albumin, triglyceride and prothrombin activity. The predictive model constructed performed well using receiver operating characteristic curve and area under the curve value, C-index and calibration curve.ConclusionThis study developed and verified a new prediction instrument for the risk of technique failure among peritoneal dialysis patients. imageConclusionThis study developed and verified a new prediction instrument for the risk of technique failure among peritoneal dialysis patients. image Technique failure is a significant obstacle for peritoneal dialysis (PD), which may be the main reason for relatively poor PD retention rates worldwide. This study aimed to establish a prediction model in PD patients to estimate the risk of technique failure and guide clinical practice. image