Data Mining Application for Detect Impacts of Infection and Hypertension on Vascular Surgery Complications

End stage renal disease (ESRD) patients need hemodialysis (HD) treatment for cleansing wastes from their blood. Vascular access (VA) – such as a lifeline - provides transporting blood from the HD patient’s vascular system to the dialyzer machine and then returning the dialyzed blood back to the patient. Arteriovenous fistula (AVF) is the first choice of the VA and is considered as the gold standard or the ideal model for VA in HD. Construction of an AVF increases the flow of blood through the venous system. Despite the popularity and wide-spread use of AVFs in the world, some complications maximize dysfunction of AVF, which, in turn, cause AVF non-maturation. Detect factors that fail VA surgery can reduces repeating surgeries and hospitalization and so improve patient ’s quality of life (QoL).Present study analyzes 484 Iranian HD patients who underwent AVF surgery from 2010 to 2017 and aged 10-90 years old. The common AVF complication between them is Thrombosis with AVF failure rate=8.85% which is less than average of reported AVF failure rates in other countries.Detected results show that the gender of ESRD patients is not directly associated with their vascular access surgery (VAS) result (Spearman s rho=-0.033, P-value=0.566). Infection has significant relation with Thrombosis (Spearman s rho=0.167, P-value<0.005), so it is recommended for patient’s infection treatment before VAS . But Hypertension affects inversely on Thrombosis (Spearman s rho=-0.160, P-value≤0.005), which means this fact that: Hypertensive HD patients have more maturated AVF . Fulfilment of our analyzes confirms both of mentioned statistical outputs, which designed by data mining approaches, such that in infection rooted in first decision tree (DT) and the rate of Thrombosis in hypertensive ESRD patients become to 5.53% which is less than in case of non-hypertensive ESRD patients. i.e. AVF failure=15.09%. The extracted knowledge obtained from DTs and the accuracy rate of classifications are up to 92.24%.

Keywords:

Decision Tree , Clustering , End Stage Renal Disease (ESRD) , Arteriovenous Fistula (AVF) , Thrombosis