Evaluation of factors associated with the success of creating the first arteriovenous system associated with SUS patients (Brazil) using machine learning techniques

The success of arteriovenous fistula (AVF) maturation in hemodialysis patients is a critical concern, with a high incidence of non-maturation or primary failure. Objective: To evaluate, using machine learning techniques, which modifiable factors influence the success of AVF maturation. This prospective, observational cohort study included all patients referred for arteriovenous fistula creation at a dialysis center in Joinville, Brazil, from January to July 2021. One hundred fifty patients were evaluated using a decision tree model with R software, physiological and social parameters, clinical data, and treatment types. Of the 150 participating patients, those of mixed race had excellent success in AVF maturation, regardless of other parameters. On the other hand, maturation success in white and black individuals depends on several other factors, such as blood pressure, obesity, and the type of treatment. Women under 58 in this ethnic group had the lowest success rate. The most important factor for this success was the site of creation, with the arm being the most important variable, followed by the type of artery, creatinine level, and lowest and highest blood pressure, as well as the type of treatment, which are associated with various physiological variables, such as BMI, patient gender, and age.