Conventional magnet manufacturing is a significant bottleneck in the development processes of products that use magnets, because every design adaption requires production steps with long lead times. Additive manufacturing of magnetic components delivers the opportunity to shift to agile and test-driven development in early prototyping stages, as well as new possibilities for complex designs. In an effort to simplify integration of magnetic components, the current work presents a method to directly print polymer-bonded hard magnets of arbitrary shape into thermoplastic parts by fused deposition modeling. This method was applied to an early prototype design of a rotary blood pump with magnetic bearing and magnetic drive coupling. Thermoplastics were compounded with 56 vol.% isotropic NdFeB powder to manufacture printable filament. With a powder loading of 56 vol.%, remanences of 350 mT and adequate mechanical flexibility for robust processability were achieved. This compound allowed us to print a prototype of a turbodynamic pump with integrated magnets in the impeller and housing in one piece on a low-cost, end-user 3D printer. Then, the magnetic components in the printed pump were fully magnetized in a pulsed Bitter coil. The pump impeller is driven by magnetic coupling to non-printed permanent magnets rotated by a brushless DC motor, resulting in a flow rate of 3 L/min at 1000 rpm. For the first time, an application of combined multi-material and magnet printing by fused deposition modeling was shown. The presented process significantly simplifies the prototyping of products that use magnets, such as rotary blood pumps, and opens the door for more complex and innovative designs. It will also help postpone the shift to conventional manufacturing methods to later phases of the development process.

Objective: The hydraulic properties of implantable rotary blood pumps (RBPs) determine their interaction with the cardiovascular system. A systematic comparison in this regard has not yet been performed for different clinically used RBPs. Aim of this study was to describe the hydraulic characteristics of four RBPs with a universal mathematical model and to compare their behavior under clinical operating conditions. Methods: First, static and dynamic pump properties of four RBPs (HVAD, Heartmate II, Heartmate 3 and Incor) including their peripheral components were identified in an in-vitro setup; results were translated into mathematical models based on principles of turbomachinery including the low and backflow region. Second, the four hydraulic models were compared in a numerical simulation of the cardiovascular system for full and partial support conditions. Results: A model structure applicable to each of the investigated RBPs was developed. Deviations between simulated and measured signals for static and dynamic properties were small (2.6 $\pm$ 0.5 mmHg, 0.38 $\pm$ 0.14 L/min, respectively). For a simulated partial support condition, flow pulsatility ranged from 4.1 (Incor) to 9.1 L/min (HVAD). Negative flow rates during diastole were observed in 3 out of 4 pumps. Conclusion: Hydraulic properties differ greatly between the investigated RBPs, with flat characteristics for the HVAD and Heartmate II and steeper curves for the Heartmate 3 and especially the Incor. Significance: Hydraulic characteristics of implantable RBPs are particularly important at lower pump flow rates if backflow is to be avoided. For further research we provide dynamic hydraulic models of the four RBPs including their periphery.

Objective: We present a novel machine learning model to accurately predict the blood-analog viscosity during support of a pathological circulation with a rotary ventricular assist device (VAD). The aim is the continuous monitoring of the hematocrit (HCT) of VAD patients with the benefit of a more reliable pump flow estimation and a possible early detection of adverse events, such as bleeding or pump thrombosis. Methods: A large dataset was generated with a blood pump connected to a hybrid mock circulation by varying the pump speed, the physiological requirements of the modelled circulation, and the viscosity of the blood-analog. The inlet pressure and the intrinsic signals of the pump were considered as inputs for the model. Gaussian process yielded models with the best performance, which were then combined using a variant of stacked generalization to derive the final model. The final model was evaluated with unseen testing data from the data set created. Results: For these data, the model yielded a mean absolute deviation of 1.81% from the true HCT, while it proved to correctly predict the direction of the HCT change. It showed to be independent of the set speed and of the condition of the simulated cardiovascular circulation. Conclusion: The accuracy of the prediction model allows an improvement of the quality of flow estimators and the detection of adverse events at an early stage. The evaluation of this approach with blood is suggested for further validation. Significance: Its clinical application could provide the clinicians with reliable and important hemodynamic information of the patient and, thus, enhance patient monitoring and supervision.