A simulation study for the design of a control system for the blood concentration process in autotransfusion.

Autotransfusion is the process in which a patient serves as his or her own blood donor; its most important application is the intraoperative blood salvage, in which the blood collected during a surgical operation is filtered, concentrated, washed, and then reinfused. In an automatic autotransfusion device, such as the DIDECO Compact Advanced, red blood cells (RBCs) are separated from the other unwanted components and concentrated by using a rotating bowl and the effect of centrifugal force. An important characteristic of concentrated RBCs is their hematocrit (Hct), i.e., percent RBC volume divided by total blood volume. The aim of this study is to assess the feasibility of a controller, based on the artificial neural network approach, which is able to provide a closed loop control of the hematocrit of the blood in the bowl at the end of the concentration phase. A simulation approach was adopted both for training the network and for assessing its performances. The results obtained are quite satisfactory, since the target Hct was typically reached within a 3% error, and always within 6% in highly challenging situations.