Prediction of equilibrated postdialysis BUN by an artificial neural network in high-efficiency hemodialysis

In urea kinetic modeling, postdialysis blood urea nitrogen (BUN) is usually underestimated with an overestimation of the Kt/V especially in high-efficiency hemodialysis (HD). Thus, an artificial neural network (ANN) was used to predict the equilibrated BUN (Ceq) and equilibrated Kt/V (eKt/V60) by using both predialysis, postdialysis, and low-flow postdialysis BUN. The results were compared to a Smye formula to predict Ceq and a Daugirdas' formula (eKt/V30) to predict eKt/V60. Seventy-four patients on high-efficiency or high-flux HD were recruited. Their mean urea rebound was 28.6+/-2%. Patients were divided into a "training" set (n = 40) and a validation set (n = 34) for the ANN. Their status was exchanged later, and the two results were pooled. In the prediction of Ceq, both Smye formula and low-flow ANN were equally highly accurate. In patients with a high urea rebound (>30%), although Smye formula lost its accuracy, low-flow ANN remained accurate. In the prediction of eKt/V60, both Daugirdas' formula and low-flow ANN were equally accurate, although the Smye formula was not so accurate. In patients with a high urea rebound, although both Smye and Daugirdas' formulas lost their accuracy, low-flow ANN remained accurate. We concluded that low-flow ANN can accurately predict both Ceq and eKt/V60 regardless of the degree of urea rebound.