Volume 62, Issue 2 - Special Issue: Smart Life Support, Issue Editor: Steffen Leonhardt

Computational models of biophysical systems generally constitute an essential component in the realization of smart biomedical technological applications. Typically, the development process of such models is characterized by a great extent of collaboration between different interdisciplinary parties. Furthermore, due to the fact that many underlying mechanisms and the necessary degree of abstraction of biophysical system models are unknown beforehand, the steps of the development process of the application are iteratively repeated when the model is refined. This paper presents some methods and tools to facilitate the development process. First, the principle of object-oriented (OO) modeling is presented and the advantages over classical signal-oriented modeling are emphasized. Second, our self-developed simulation tool ModeliChart is presented. ModeliChart was designed specifically for clinical users and allows independently performing in silico studies in real time including intuitive interaction with the model. Furthermore, ModeliChart is capable of interacting with hardware such as sensors and actuators. Finally, it is presented how optimal control methods in combination with OO models can be used to realize clinically motivated control applications. All methods presented are illustrated on an exemplary clinically oriented use case of the artificial perfusion of the systemic circulation.

Background: Left ventricular assist device (LVAD) therapy has revolutionized the way end stage heart failure is treated today. Analysis of LVAD interaction with the whole cardiovascular system and its biological feedback loops is often conducted by means of computer models. Generating real time pressure volume loops (PV-loops) in patients, not using conductance catheters but routine diagnostics to feed an in-silico model could help to predict postoperative complications. Methods: Routinely obtained hemodynamic measurements to evaluate myocardial function prior to LVAD implantation like pressure readings in the aorta, the left atrium and the left ventricle and simultaneous three-dimensional (3D) echocardiography recordings were assessed to parametrize a reduced computational model of the cardiovascular system. An automatic parameter identification procedure has been developed. Results: The results constitute a patient-individual computational simulation model. An exemplary in-silico study focusing on the effect of different ventricular assist device (VAD) speeds has been conducted. Results allow for estimation of the resulting hemodynamic parameters and changes of the PV-loops. Conclusion: The model improves understanding and prediction of the interaction between pump and ventricles. Future modifications in exporting and merging routinely assessed real time hemodynamic patient data are necessary to investigate various clinical and pathological conditions of LVAD recipients.

Right heart failure (RHF), e.g. due to pulmonary hypertension (PH), is a serious health issue with growing occurrence and high mortality rate. Limited efficacy of medication in advanced stages of the disease constitutes the need for mechanical circulatory support of the right ventricle (RV). An essential contribution to the process of developing right ventricular assist devices (RVADs) is the in vitro test bench, which simulates the hemodynamic behavior of the native circulatory system. To model healthy and diseased arterial-pulmonary hemodynamics in adults (mild and severe PH and RHF), a right heart mock circulation loop (MCL) was developed. Incorporating an anatomically shaped silicone RV and a silicone atrium, it not only enables investigations of hemodynamic values but also suction events or the handling of minimal invasive RVADs in an anatomical test environment. Ventricular pressure-volume loops of all simulated conditions as well as pressure and volume waveforms were recorded and compared to literature data. In an exemplary test, an RVAD was connected to the apex to further test the feasibility of studying such devices with the developed MCL. In conclusion, the hemodynamic behavior of the native system was well reproduced by the developed MCL, which is a useful basis for future RVAD tests.

Mock heart circulation loops (MHCLs) serve as in-vitro platforms to investigate the physiological interaction between circulatory systems and cardiovascular devices. A mock heart (MH) engineered with silicone walls and helical aramid fibers, to mimic the complex contraction of a natural heart, has been developed to advance the MHCL previously developed in our group. A mock aorta with an anatomical shape enables the evaluation of a cannulation method for ventricular assist devices (VADs) and investigation of the usage of clinical measurement systems like pressure-volume catheters. Ventricle and aorta molds were produced based on MRI data and cast with silicone. Aramid fibers were layered in the silicone ventricle to reproduce ventricle torsion. A rotating hollow shaft was connected to the apex enabling the rotation of the MH and the connection of a VAD. Silicone wall thickness, aramid fiber angle and fiber pitch were varied to generate different MH models. All MH models were placed in a tank filled with variable amounts of water and air simulating the compliance. In this work, physiological ventricular torsion angles (15°–26°) and physiological pressure-volume loops were achieved. This MHCL can serve as a comprehensive testing platform for cardiovascular devices, such as artificial heart valves and cannulation of VADs.

Terminal heart failure (HF) is the most prevalent cause of death in the Western world and the implantation of a left ventricular assist device (LVAD) has become the gold standard therapy today. Most of the actually implanted devices are driven at a constant speed (CS) regardless of the patient’s physiological demand. A new physiological controller [power ratio (PR) controller], which keeps a constant ratio between LVAD power and left ventricular power, a previous concept [preload responsive speed (PRS) controller], which adds a variable LVAD power to reach a defined stroke work, and a CS controller were compared with an unimpaired ventricle in a full heart computer simulation model. The effects of changes in preload, afterload and left ventricular contractility are displayed by global hemodynamics and ventricular pressure-volume loops. Both physiological controllers demonstrated the desired load dependency, whereas the PR controller exceeded the PRS controller in response to an increased load and contractility. Response was inferior when preload or contractility was decreased. Thus, the PR controller might lead to an increased exercise tolerance of the patient. Additional studies are required to evaluate the controllers in vivo .

Ventricular assist devices (VADs) are mechanical blood pumps that are clinically used to treat severe heart failure. Pulsatile VADs (pVADs) were initially used, but are today in most cases replaced by turbodynamic VADs (tVADs). The major concern with the pVADs is their size, which prohibits full pump body implantation for a majority of patients. A reduction of the necessary stroke volume can be achieved by increasing the stroke frequency, while maintaining the same level of support capability. This reduction in stroke volume in turn offers the possibility to reduce the pump’s overall dimensions. We simulated a human cardiovascular system (CVS) supported by a pVAD with three different stroke rates that were equal, two- or threefold the heart rate (HR). The pVAD was additionally synchronized to the HR for better control over the hemodynamics and the ventricular unloading. The simulation results with a HR of 90 bpm showed that a pVAD stroke volume can be reduced by 71%, while maintaining an aortic pulse pressure (PP) of 30 mm Hg, avoiding suction events, reducing the ventricular stroke work (SW) and allowing the aortic valve to open. A reduction by 67% offers the additional possibility to tune the interaction between the pVAD and the CVS. These findings allow a major reduction of the pVAD’s body size, while allowing the physician to tune the pVAD according to the patient’s needs.

Electromyography (EMG) has long been used for the assessment of muscle function and activity and has recently been applied to the control of medical ventilation. For this application, the EMG signal is usually recorded invasively by means of electrodes on a nasogastric tube which is placed inside the esophagus in order to minimize noise and crosstalk from other muscles. Replacing these invasive measurements with an EMG signal obtained non-invasively on the body surface is difficult and requires techniques for signal separation in order to reconstruct the contributions of the individual respiratory muscles. In the case of muscles with small cross-sectional areas, or with muscles at large distances from the recording site, solutions to this problem have been proposed previously. The respiratory muscles, however, are large and distributed widely over the upper body volume. In this article, we describe an algorithm for convolutive blind source separation (BSS) that performs well even for large, distributed muscles such as the respiratory muscles, while using only a small number of electrodes. The algorithm is derived as a special case of the TRINICON general framework for BSS. To provide evidence that it shows potential for separating inspiratory, expiratory, and cardiac activities in practical applications, a joint numerical simulation of EMG and ECG activities was performed, and separation success was evaluated in a variety of noise settings. The results are promising.

Mathematical physiological models can be applied in medical decision support systems. To do so requires consideration of the necessary model complexity. Models that simulate changes in the individual patient are required, meaning that models should have a complexity where parameters can be uniquely identified at the bedside from clinical data and where the models adequately represent the individual patient’s (patho)physiology. This paper describes the models included in a system for providing decision support for mechanical ventilation. Models of pulmonary gas exchange, respiratory mechanics, acid-base, and respiratory control are described. The parameters of these models are presented along with the necessary clinical data required for their estimation and the parameter estimation process. In doing so, the paper highlights the need for simple, minimal models for application at the bedside, directed toward well-defined clinical problems.

A new concept is presented for cooperative automation of mechanical ventilation and extracorporeal membrane oxygenation (ECMO) therapy for treatment of acute respiratory distress syndrome (ARDS). While mechanical ventilation is continuously optimized to promote lung protection, extracorporeal gas transfer rates are simultaneously adjusted to control oxygen supply and carbon dioxide removal using a robust patient-in-the-loop control system. In addition, the cooperative therapy management uses higher-level algorithms to adjust both therapeutic approaches. The controller synthesis is derived based on the introduced objectives, the experimental setup and the uncertain models. Finally, the autonomous ARDS therapy system capabilities are demonstrated and discussed based on in vivo data from animal experiments.

Abstract: This paper presents a decentralized safety concept for networked intensive care setups, for which a decentralized network of sensors and actuators is realized by embedded microcontroller nodes. It is evaluated for up to eleven medical devices in a setup for automated acute respiratory distress syndrome (ARDS) therapy. In this contribution we highlight a blood pump supervision as exemplary safety measure, which allows a reliable bubble detection in an extracorporeal blood circulation. The approach is validated with data of animal experiments including 35 bubbles with a size between 0.05 and 0.3 ml. All 18 bubbles with a size down to 0.15 ml are successfully detected. By using hidden Markov models (HMMs) as statistical method the number of necessary sensors can be reduced by two pressure sensors.

Very/extremely premature infants often experience glycaemic dysregulation, resulting in abnormally elevated (hyperglycaemia) or low (hypoglycaemia) blood glucose (BG) concentrations, due to prematurity, stress, and illness. STAR-GRYPHON is a computerised protocol that utilises a model-based insulin sensitivity parameter to directly tailor therapy for individual patients and their changing conditions, unlike other common insulin protocols in this cohort. From January 2013 to January 2015, 13 patients totalling 16 hyperglycaemic control episodes received insulin under STAR-GRYPHON. A significant improvement in control was achieved in comparison to a retrospective cohort, with a 26% absolute improvement in BG within the targeted range and no hypoglycaemia. This improvement was obtained predominantly due to the reduction of hyperglycaemia (%BG>10.0 mmol/l: 5.6 vs. 17.7%, p<0.001), and lowering of the median per-patient BG [6.9 (6.1–7.9) vs. 7.8 (6.6–9.1) mmol/l, p<0.001, Mann-Witney U test]. While cohort-wide control results show good control overall, there is high intra-patient variability in BG behaviour, resulting in overly conservative treatments for some patients. Patient insulin sensitivity differs between and within patients over time, with some patients having stable insulin sensitivity, while others change rapidly. These results demonstrate the trade-off between safety and performance in a highly variable and fragile cohort.