Abstract: The rising prevalence of end-stage renal disease (ESRD) promotes the application of peritoneal dialysis (PD), a therapeutic option to conveniently conduct at home. However, PD-associated infectious and non-infectious complications occur frequently. Early and precise prediction and management of PD-associated complications are therefore critical to quality of life and outcome of the patients. The risk prediction models based on machine learning can evaluate the complication risk of a patient; while its branch, the deep machine learning, can automatically extract hierarchical features from raw data with higher accuracy and efficiency through multi-layer neural network structures. This review summarizes the basic concepts of machine learning and its branch deep machine learning, their utility in PD complication prediction and recent research advances to provide references for complication management in PD patients.

Key words: Machine learning, Peritoneal dialysis, Risk prediction model