Explainable Machine Learning Based Prediction of Severity of Heart Failure Using Primary Electronic Health Records.

Heart Failure (HF) is a life-threatening condition. It affects more than 64 million people worldwide. Early diagnosis of HF is extremely crucial. In this study, we propose utilization of machine learning (ML) models to predict severity of HF from primary Electronic Health Records (EHRs). We used a public dataset of 2008 HF patients for the study. Gaussian Naive Bayes, Random Forest and CatBoost methods were used for prediction. The study shows that CatBoost works best for the goal. In addition to that, the largest contributors for tree-based models harmonize well with clinically important parameters, which exhibits the trustworthiness of these models. Hence, we conclude that utilization of ML methods on primary EHRs is a promising step for time-efficient diagnosis of HF patients.

Computational modelling of cardiovascular pathophysiology to risk stratify commercial spaceflight.

For more than 60 years, humans have travelled into space. Until now, the majority of astronauts have been professional, government agency astronauts selected, in part, for their superlative physical fitness and the absence of disease. Commercial spaceflight is now becoming accessible to members of the public, many of whom would previously have been excluded owing to unsatisfactory fitness or the presence of cardiorespiratory diseases. While data exist on the effects of gravitational and acceleration (G) forces on human physiology, data on the effects of the aerospace environment in unselected members of the public, and particularly in those with clinically significant pathology, are limited. Although short in duration, these high acceleration forces can potentially either impair the experience or, more seriously, pose a risk to health in some individuals. Rather than expose individuals with existing pathology to G forces to collect data, computational modelling might be useful to predict the nature and severity of cardiovascular diseases that are of sufficient risk to restrict access, require modification, or suggest further investigation or training before flight. In this Review, we explore state-of-the-art, zero-dimensional, compartmentalized models of human cardiovascular pathophysiology that can be used to simulate the effects of acceleration forces, homeostatic regulation and ventilation-perfusion matching, using data generated by long-arm centrifuge facilities of the US National Aeronautics and Space Administration and the European Space Agency to risk stratify individuals and help to improve safety in commercial suborbital spaceflight.

Strain-controlled electrophysiological wave propagation alters in silico scar-based substrate for ventricular tachycardia.

Introduction: Assessing a patient's risk of scar-based ventricular tachycardia (VT) after myocardial infarction is a challenging task. It can take months to years after infarction for VT to occur. Also, if selected for ablation therapy, success rates are low. Methods: Computational ventricular models have been presented previously to support VT risk assessment and to provide ablation guidance. In this study, an extension to such virtual-heart models is proposed to phenomenologically incorporate tissue remodeling driven by mechanical load. Strain amplitudes in the heart muscle are obtained from simulations of mechanics and are used to adjust the electrical conductivity. Results: The mechanics-driven adaptation of electrophysiology resulted in a more heterogeneous distribution of propagation velocities than that of standard models, which adapt electrophysiology in the structural substrate from medical images only. Moreover, conduction slowing was not only present in such a structural substrate, but extended in the adjacent functional border zone with impaired mechanics. This enlarged the volumes with high repolarization time gradients (≥10 ms/mm). However, maximum gradient values were not significantly affected. The enlarged volumes were localized along the structural substrate border, which lengthened the line of conduction block. The prolonged reentry pathways together with conduction slowing in functional regions increased VT cycle time, such that VT was easier to induce, and the number of recommended ablation sites increased from 3 to 5 locations. Discussion: Sensitivity testing showed an accurate model of strain-dependency to be critical for low ranges of conductivity. The model extension with mechanics-driven tissue remodeling is a potential approach to capture the evolution of the functional substrate and may offer insight into the progression of VT risk over time.

Efficient sensitivity analysis for biomechanical models with correlated inputs.

In most variance-based sensitivity analysis (SA) approaches applied to biomechanical models, statistical independence of the model input is assumed. However, often the model inputs are correlated. This might alter the interpretation of the SA results, which may severely impact the guidance provided during model development and personalization. Potential reasons for the infrequent usage of SA techniques that account for input correlation are the associated high computational costs, especially for models with many parameters, and the fact that the input correlation structure is often unknown. The aim of this study was to propose an efficient correlated global sensitivity analysis method by applying a surrogate model-based approach. Furthermore, this article demonstrates how correlated SA should be interpreted and how the applied method can guide the modeler during model development and personalization, even when the correlation structure is not entirely known beforehand. The proposed methodology was applied to a typical example of a pulse wave propagation model and resulted in accurate SA results that could be obtained at a theoretically 27,000× lower computational cost compared to the correlated SA approach without employing a surrogate model. Furthermore, our results demonstrate that input correlations can significantly affect SA results, which emphasizes the need to thoroughly investigate the effect of input correlations during model development. We conclude that our proposed surrogate-based SA approach allows modelers to efficiently perform correlated SA to complex biomechanical models and allows modelers to focus on input prioritization, input fixing and model reduction, or assessing the dependency structure between parameters.

Wearable technology and the cardiovascular system: the future of patient assessment.

The past decade has seen a dramatic rise in consumer technologies able to monitor a variety of cardiovascular parameters. Such devices initially recorded markers of exercise, but now include physiological and health-care focused measurements. The public are keen to adopt these devices in the belief that they are useful to identify and monitor cardiovascular disease. Clinicians are therefore often presented with health app data accompanied by a diverse range of concerns and queries. Herein, we assess whether these devices are accurate, their outputs validated, and whether they are suitable for professionals to make management decisions. We review underpinning methods and technologies and explore the evidence supporting the use of these devices as diagnostic and monitoring tools in hypertension, arrhythmia, heart failure, coronary artery disease, pulmonary hypertension, and valvular heart disease. Used correctly, they might improve health care and support research.

Deep Learning Method for Estimation of Morphological Parameters Based on CT Scans.

In this study, we propose a Convolutional Neural Network (CNN) with an assembly of non-linear fully connected layers for estimating body height and weight using a limited amount of data. This method can predict the parameters within acceptable clinical limits for most of the cases even when trained with limited data.

Evaluating the Arteriotomy Size of a New Sutureless Coronary Anastomosis Using a Finite Volume Approach.

OBJECTIVES: The ELANA® Heart Bypass creates a standardized sutureless anastomosis. Hereby, we investigate the influence of arteriotomy and graft size on coronary hemodynamics. METHODS: A computational fluid dynamics (CFD) model was developed. Arteriotomy size (standard 1.43 mm2; varied 0.94 - 3.6 mm2) and graft diameter (standard 2.5 mm; varied 1.5 - 5.0 mm) were independent parameters. Outcome parameters were coronary pressure and flow, and fractional flow reserve (FFR). RESULTS: The current size ELANA (arteriotomy 1.43 mm2) presented an estimated FFR 0.65 (39 mL/min). Enlarging arteriotomy increased FFR, coronary pressure, and flow. All reached a maximum once the arteriotomy (2.80 mm2) surpassed the coronary cross-sectional area (2.69 mm2, i.e. 1.85 mm diameter), presenting an estimated FFR 0.75 (46 mL/min). Increasing graft diameter was positively related to FFR, coronary pressure, and flow. CONCLUSION: The ratio between the required minimal coronary diameter for application and the ELANA arteriotomy size effectuates a pressure drop that could be clinically relevant. Additional research and eventual lengthening of the anastomosis is advised.

MarioHeart: Novel In-Vitro Flow Model for Testing Heart Valve Prostheses and Anticoagulant Therapies.

Mechanical heart valve (MHV) prostheses present a risk of thromboembolic complications despite antithrombotic therapy. Further steps in the development of more hemocompatible MHVs and new anticoagulants are impeded due to the lack of adequate in-vitro models. With the development of a novel in-vitro model (MarioHeart), a pulsatile flow similar to the arterial circulation is emulated. The MarioHeart design owns unique features as 1) a single MHV within a torus with low surface/volume ratio, 2) a closed loop system, and 3) a dedicated external control system driving the oscillating rotational motion of the torus. For verification purposes, a blood analog fluid seeded with particles was used to assess fluid velocity and flow rate using a speckle tracking method on high-speed video recordings of the rotating model. The flow rate resembled the physiological flow rate in the aortic root, in both shape and amplitude. Additional in-vitro runs with porcine blood showed thrombi on the MHV associated with the suture ring, which is similar to the in-vivo situation. MarioHeart is a simple design which induces well-defined fluid dynamics resulting in physiologically nonturbulent flow without stasis of the blood. MarioHeart seems suitable for testing the thrombogenicity of MHVs and the potential of new anticoagulants.

Spatiotemporal Registration of 3-D Multi-perspective Ultrasound Images of Abdominal Aortic Aneurysms.

Methods for patient-specific abdominal aortic aneurysm (AAA) progression monitoring and rupture risk assessment are widely investigated. Three-dimensional ultrasound can visualize the AAA's complex geometry and displacement fields. However, ultrasound has a limited field of view and low frame rate (i.e., 3-8 Hz). This article describes an approach to enhance the temporal resolution and the field of view. First, the frame rate was increased for each data set by sequencing multiple blood pulse cycles into one cycle. The sequencing method uses the original frame rate and the estimated pulse wave rate obtained from AAA distension curves. Second, the temporal registration was applied to multi-perspective acquisitions of the same AAA. Third, the field of view was increased through spatial registration and fusion using an image feature-based phase-only correlation method and a wavelet transform, respectively. Temporal sequencing was fully correct in aortic phantoms and was successful in 51 of 62 AAA patients, yielding a factor 5 frame rate increase. Spatial registration of proximal and distal ultrasound acquisitions was successful in 32 of 37 different AAA patients, based on the comparison between the fused ultrasound and computed tomography segmentation (95th percentile Haussdorf distances and similarity indices of 4.2 ± 1.7 mm and 0.92 ± 0.02 mm, respectively). Furthermore, the field of view was enlarged by 9%-49%.

A review study of fetal circulatory models to develop a digital twin of a fetus in a perinatal life support system.

Background: Preterm birth is the main cause of neonatal deaths with increasing mortality and morbidity rates with decreasing GA at time of birth. Currently, premature infants are treated in neonatal intensive care units to support further development. However, the organs of, especially, extremely premature infants (born before 28 weeks of GA) are not mature enough to function optimally outside the womb. This is seen as the main cause of the high morbidity and mortality rates in this group. A liquid-filled incubator, a so-called PLS system, could potentially improve these numbers for extremely premature infants, since this system is designed to mimic the environment of the natural womb. To support the development and implementation of such a complex system and to interpret vital signals of the fetus during a PLS system operation, a digital twin is proposed. This mathematical model is connected with a manikin representing the digital and physical twin of the real-life PLS system. Before developing a digital twin of a fetus in a PLS system, its functional and technical requirements are defined and existing mathematical models are evaluated. Method and results: This review summarizes existing 0D and 1D fetal circulatory models that potentially could be (partly) adopted for integration in a digital twin of a fetus in a PLS system based on predefined requirements. The 0D models typically describe hemodynamics and/or oxygen transport during specific events, such as the transition from fetus to neonate. Furthermore, these models can be used to find hemodynamic differences between healthy and pathological physiological states. Rather than giving a global description of an entire cardiovascular system, some studies focus on specific organs or vessels. In order to analyze pressure and flow wave profiles in the cardiovascular system, transmission line or 1D models are used. As for now, these models do not include oxygen transport. Conclusion: This study shows that none of the models identified in literature meet all the requirements relevant for a digital twin of a fetus in a PLS system. Nevertheless, it does show the potential to develop this digital twin by integrating (parts) of models into a single model.

From evidence-based medicine to digital twin technology for predicting ventricular tachycardia in ischaemic cardiomyopathy.

Survivors of myocardial infarction are at risk of life-threatening ventricular tachycardias (VTs) later in their lives. Current guidelines for implantable cardioverter defibrillators (ICDs) implantation to prevent VT-related sudden cardiac death is solely based on symptoms and left ventricular ejection fraction. Catheter ablation of scar-related VTs is performed following ICD therapy, reducing VTs, painful shocks, anxiety, depression and worsening heart failure. We postulate that better prediction of the occurrence and circuit of VT, will improve identification of patients at risk for VT and boost preventive ablation, reducing mortality and morbidity. For this purpose, multiple time-evolving aspects of the underlying pathophysiology, including the anatomical substrate, triggers and modulators, should be part of VT prediction models. We envision digital twins as a solution combining clinical expertise with three prediction approaches: evidence-based medicine (clinical practice), data-driven models (data science) and mechanistic models (biomedical engineering). This paper aims to create a mutual understanding between experts in the different fields by providing a comprehensive description of the clinical problem and the three approaches in an understandable manner, leveraging future collaborations and technological innovations for clinical decision support. Moreover, it defines open challenges and gains for digital twin solutions and discusses the potential of hybrid modelling.

Corrigendum: Ultrasound-Based Fluid-Structure Interaction Modeling of Abdominal Aortic Aneurysms Incorporating Pre-Stress.

[This corrects the article DOI: 10.3389/fphys.2021.717593.].

A strategy to personalize a 1D pulse wave propagation model for estimating subject-specific central aortic pressure waveform.

The central aortic pressure (CAP) provides insights into the prediction, prevention, diagnosis, and treatment of cardiovascular disease, but can't be directly measured non-invasively. Therefore, the development of a noninvasive CAP estimation method based on the non-invasively measured peripheral pressure waveform is critical for clinical decisions based on the CAP. Some existing widely applied methods, such as the generalized transfer function (GTF) method relating measured peripheral pressure to the CAP, do not or only partly account for inter-subject or intra-subject variability of the cardiovascular system. To overcome this pitfall, we propose a subject-specific central aortic pressure estimation method in this paper. The novel method presented can derive an accurate aortic pressure from the peripheral pressure based on an individualized pulse wave propagation model using a GTF method as a first guess. To develop a strategy to personalize a pulse wave propagation model one usually needs to optimize many input parameters. Therefore, we present a two-step approach with the screening method of Morris and the adaptive sparse generalized polynomial chaos expansion (agPCE) algorithm for the sensitivity analysis of the wave propagation model. First, for a-priori defined output of the model, a subset of important parameters is identified using the screening method of Morris. Next, a quantitative variance-based sensitivity analysis is performed using agPCE. This approach is applied to a 1D pulse wave propagation model to get the personalized parameters of the pulse wave propagation model for the estimation of a subject-specific central aortic pressure waveform and is validated with 26 patients. Compared with the GTF method, the proposed method showed better performance in estimating the central aortic pulse wave and predicting the parameters.

Continuum modeling of thrombus formation and growth under different shear rates.

Obstruction of blood flow due to thrombosis is a major cause of ischemic stroke, myocardial infarction, and in severe cases, mortality. In particular, in blood wetting medical devices, thrombosis is a common reason for failure. The prediction of thrombosis by understanding signaling pathways using computational models, lead to identify the risk of thrombus formation in blood-contacting devices and design improvements. In this study, a mathematical model of thrombus formation and growth is presented. A biochemical model of platelet activation and aggregation is developed to predict thrombus size and shape at the site of vascular injury. Computational fluid dynamics using the finite volume method is employed to compute the velocity and pressure fields which are influenced by the growing thrombi. The passive transport of platelets, agonists, the platelet activation kinetics, their adhesion to the growing thrombi and embolization of platelets are solved by a fully coupled set of convection-diffusion-reaction equations. The thrombogenic surface representing blood-contacting material or injured blood vessel was incorporated into the model as a surface flux boundary condition to initiate thrombus formation. The blood is considered as a Newtonian fluid, while the thrombus is treated as a porous medium. The results are compared with in vitro experiments of a microfluidic chamber at an initial inlet venous shear rate of 200s-1 using a pressure-inlet boundary condition. The thrombus development due to agonist concentrations and change in the shear rate as well as thromboembolism for this benchmark problem is successfully computed. Furthermore, to extend the current model to a physiologically relevant configuration, thrombus formation in a blood tube is simulated. Two different heterogeneous reaction rates for platelet aggregation are used to simulate thrombus growth under a constant inlet flow rate. The findings show that the thrombus shape can be substantially altered by the hemodynamic conditions, increase in the shear rate and due to the combined effects of shear induced platelet activation and the heterogeneous reaction rates. It is also concluded that the model is able to predict thrombus formation in different physiological and pathological hemodynamics.

Local and global distensibility assessment of abdominal aortic aneurysms in vivo from probe tracked 2D ultrasound images.

Rupture risk estimation of abdominal aortic aneurysm (AAA) patients is currently based on the maximum diameter of the AAA. Mechanical properties that characterize the mechanical state of the vessel may serve as a better rupture risk predictor. Non-electrocardiogram-gated (non-ECG-gated) freehand 2D ultrasound imaging is a fast approach from which a reconstructed volumetric image of the aorta can be obtained. From this 3D image, the geometry, volume, and maximum diameter can be obtained. The distortion caused by the pulsatility of the vessel during the acquisition is usually neglected, while it could provide additional quantitative parameters of the vessel wall. In this study, a framework was established to semi-automatically segment probe tracked images of healthy aortas (N = 10) and AAAs (N = 16), after which patient-specific geometries of the vessel at end diastole (ED), end systole (ES), and at the mean arterial pressure (MAP) state were automatically assessed using heart frequency detection and envelope detection. After registration AAA geometries were compared to the gold standard computed tomography (CT). Local mechanical properties, i.e., compliance, distensibility and circumferential strain, were computed from the assessed ED and ES geometries for healthy aortas and AAAs, and by using measured brachial pulse pressure values. Globally, volume, compliance, and distensibility were computed. Geometries were in good agreement with CT geometries, with a median similarity index and interquartile range of 0.91 [0.90-0.92] and mean Hausdorff distance and interquartile range of 4.7 [3.9-5.6] mm. As expected, distensibility (Healthy aortas: 80 ± 15·10-3 kPa-1; AAAs: 29 ± 9.6·10-3 kPa-1) and circumferential strain (Healthy aortas: 0.25 ± 0.03; AAAs: 0.15 ± 0.03) were larger in healthy vessels compared to AAAs. Circumferential strain values were in accordance with literature. Global healthy aorta distensibility was significantly different from AAAs, as was demonstrated with a Wilcoxon test (p-value = 2·10-5). Improved image contrast and lateral resolution could help to further improve segmentation to improve mechanical characterization. The presented work has demonstrated how besides accurate geometrical assessment freehand 2D ultrasound imaging is a promising tool for additional mechanical property characterization of AAAs.

Ultrasound-Based Fluid-Structure Interaction Modeling of Abdominal Aortic Aneurysms Incorporating Pre-stress.

Currently, the prediction of rupture risk in abdominal aortic aneurysms (AAAs) solely relies on maximum diameter. However, wall mechanics and hemodynamics have shown to provide better risk indicators. Patient-specific fluid-structure interaction (FSI) simulations based on a non-invasive image modality are required to establish a patient-specific risk indicator. In this study, a robust framework to execute FSI simulations based on time-resolved three-dimensional ultrasound (3D+t US) data was obtained and employed on a data set of 30 AAA patients. Furthermore, the effect of including a pre-stress estimation (PSE) to obtain the stresses present in the measured geometry was evaluated. The established workflow uses the patient-specific 3D+t US-based segmentation and brachial blood pressure as input to generate meshes and boundary conditions for the FSI simulations. The 3D+t US-based FSI framework was successfully employed on an extensive set of AAA patient data. Omitting the pre-stress results in increased displacements, decreased wall stresses, and deviating time-averaged wall shear stress and oscillatory shear index patterns. These results underline the importance of incorporating pre-stress in FSI simulations. After validation, the presented framework provides an important tool for personalized modeling and longitudinal studies on AAA growth and rupture risk.

A Generalized Approach for Automatic 3-D Geometry Assessment of Blood Vessels in Transverse Ultrasound Images Using Convolutional Neural Networks.

Accurate 3-D geometries of arteries and veins are important clinical data for diagnosis of arterial disease and intervention planning. Automatic segmentation of vessels in the transverse view suffers from the low lateral resolution and contrast. Convolutional neural networks are a promising tool for automatic segmentation of medical images, outperforming the traditional segmentation methods with high robustness. In this study, we aim to create a general, robust, and accurate method to segment the lumen-wall boundary of healthy central and peripheral vessels in large field-of-view freehand ultrasound (US) datasets. Data were acquired using the freehand US, in combination with a probe tracker. A total of ±36 000 cross-sectional images, acquired in the common, internal, and external carotid artery ( N = 37 ), in the radial, ulnar artery, and cephalic vein ( N = 12 ), and in the femoral artery ( N = 5 ) were included. To create masks (of the lumen) for training data, a conventional automatic segmentation method was used. The neural networks were trained on: 1) data of all vessels and 2) the carotid artery only. The performance was compared and tested using an open-access dataset. The recall, precision, DICE, and intersection over union (IoU) were calculated. Overall, segmentation was successful in the carotid and peripheral arteries. The Multires U-net architecture performs best overall with DICE = 0.93 when trained on the total dataset. Future studies will focus on the inclusion of vascular pathologies.

Model-based aortic power transfer: A potential measure for quantifying aortic stenosis severity based on measured data.

Current aortic stenosis severity grading is based mainly on the local properties of the stenotic valve, such as pressure gradient or jet velocity. Success rates of valve replacement therapy are still suboptimal, so alternative grading of AS should be investigated. We suggest the efficiency of power transfer from the left ventricle to the aorta, as it takes into account heart, valve and circulatory system. Left ventricular and circulatory power were estimated using a 0D model, which was optimised to patient data: left ventricular and aortic pressure, aortic flow and diastolic left ventricular volume. Optimisation was performed using a data assimilation method. These data were available in rest as well as chemically induced exercise for twelve patients. Using this limited data set, we showed that aortic valve efficiency is highly heterogeneous between patients, but also often dependent on the haemodynamic load. This indicates that power transfer efficiency is a highly interesting metric for further research in aortic stenosis.

Interprofessional Consensus Regarding Design Requirements for Liquid-Based Perinatal Life Support (PLS) Technology.

Liquid-based perinatal life support (PLS) technology will probably be applied in a first-in-human study within the next decade. Research and development of PLS technology should not only address technical issues, but also consider socio-ethical and legal aspects, its application area, and the corresponding design implications. This paper represents the consensus opinion of a group of healthcare professionals, designers, ethicists, researchers and patient representatives, who have expertise in tertiary obstetric and neonatal care, bio-ethics, experimental perinatal animal models for physiologic research, biomedical modeling, monitoring, and design. The aim of this paper is to provide a framework for research and development of PLS technology. These requirements are considering the possible respective user perspectives, with the aim to co-create a PLS system that facilitates physiological growth and development for extremely preterm born infants.

Corrections to "Multiperspective Ultrasound Strain Imaging of the Abdominal Aorta".

In the above article [1], one error, (1), was found which has an impact on the results and interpretation of the image quality that was described in four sentences, (2)-(5). This correction does not mitigate the overall conclusion of this work, but does positively support improvements of the proposed method (multi-perspective ultrafast imaging) in comparison with the standard method (focused line-by-line imaging).