Evolution-based Shape and Behavior Co-design of Virtual Agents.

We introduce a novel co-design method for autonomous moving agents' shape attributes and locomotion by combining deep reinforcement learning and evolution with user control. Our main inspiration comes from evolution, which has led to wide variability and adaptation in Nature and has significantly improved design and behavior simultaneously. Our method takes an input agent with optional user-defined constraints, such as leg parts that should not evolve or are only within the allowed ranges of changes. It uses physics-based simulation to determine its locomotion and finds a behavior policy for the input design that is used as a baseline for comparison. The agent is randomly modified within the allowed ranges, creating a new generation of several hundred agents. The generation is trained by transferring the previous policy, which significantly speeds up the training. The best-performing agents are selected, and a new generation is formed using their crossover and mutations. The next generations are then trained until satisfactory results are reached. We show a wide variety of evolved agents, and our results show that even with only 10 the overall performance of the evolved agents improves by 50 experiments' performance will improve even more to 150 structures, and it does not require considerable computation resources as it works on a single GPU and provides results by training thousands of agents within 30 minutes.

Shrinkage Compensation and Effect of Building Orientation on Mechanical Properties of Ceramic Stereolithography Parts.

Stereolithography additive manufacturing (SLA-AM) can be used to produce ceramic structures by selectively curing a photosensitive resin that has ceramic powder in it. The photosensitive resin acts as a ceramic powder binder, which is burned, and the remaining ceramic part is sintered during post-processing using a temperature-time-controlled furnace. Due to this process, the ceramic part shrinks and becomes porous. Moreover, additive manufacturing leads to the orthotropic behavior of the manufactured parts. This article studies the effect of the manufacturing orientation of ceramic parts produced via SLA-AM on dimensional accuracy. Scaled CAD models were created by including the calculated shrinkage factor. The dimensions of the final sintered specimens were very close to the desired dimensions. As sintering induces porosity and reduces the mechanical strength, in this study, the effect of orientation on strength was investigated, and it was concluded that the on-edge specimen possessed by far the highest strength in terms of both compression and tension.

Exploring interpretability in deep learning prediction of successful ablation therapy for atrial fibrillation.

Background: Radiofrequency catheter ablation (RFCA) therapy is the first-line treatment for atrial fibrillation (AF), the most common type of cardiac arrhythmia globally. However, the procedure currently has low success rates in dealing with persistent AF, with a reoccurrence rate of ∼50% post-ablation. Therefore, deep learning (DL) has increasingly been applied to improve RFCA treatment for AF. However, for a clinician to trust the prediction of a DL model, its decision process needs to be interpretable and have biomedical relevance. Aim: This study explores interpretability in DL prediction of successful RFCA therapy for AF and evaluates if pro-arrhythmogenic regions in the left atrium (LA) were used in its decision process. Methods: AF and its termination by RFCA have been simulated in MRI-derived 2D LA tissue models with segmented fibrotic regions (n = 187). Three ablation strategies were applied for each LA model: pulmonary vein isolation (PVI), fibrosis-based ablation (FIBRO) and a rotor-based ablation (ROTOR). The DL model was trained to predict the success of each RFCA strategy for each LA model. Three feature attribution (FA) map methods were then used to investigate interpretability of the DL model: GradCAM, Occlusions and LIME. Results: The developed DL model had an AUC (area under the receiver operating characteristic curve) of 0.78 ± 0.04 for predicting the success of the PVI strategy, 0.92 ± 0.02 for FIBRO and 0.77 ± 0.02 for ROTOR. GradCAM had the highest percentage of informative regions in the FA maps (62% for FIBRO and 71% for ROTOR) that coincided with the successful RFCA lesions known from the 2D LA simulations, but unseen by the DL model. Moreover, GradCAM had the smallest coincidence of informative regions of the FA maps with non-arrhythmogenic regions (25% for FIBRO and 27% for ROTOR). Conclusion: The most informative regions of the FA maps coincided with pro-arrhythmogenic regions, suggesting that the DL model leveraged structural features of MRI images to identify such regions and make its prediction. In the future, this technique could provide a clinician with a trustworthy decision support tool.

Imaging and biophysical modelling of thrombogenic mechanisms in atrial fibrillation and stroke.

Atrial fibrillation (AF) underlies almost one third of all ischaemic strokes, with the left atrial appendage (LAA) identified as the primary thromboembolic source. Current stroke risk stratification approaches, such as the CHA2DS2-VASc score, rely mostly on clinical comorbidities, rather than thrombogenic mechanisms such as blood stasis, hypercoagulability and endothelial dysfunction-known as Virchow's triad. While detection of AF-related thrombi is possible using established cardiac imaging techniques, such as transoesophageal echocardiography, there is a growing need to reliably assess AF-patient thrombogenicity prior to thrombus formation. Over the past decade, cardiac imaging and image-based biophysical modelling have emerged as powerful tools for reproducing the mechanisms of thrombogenesis. Clinical imaging modalities such as cardiac computed tomography, magnetic resonance and echocardiographic techniques can measure blood flow velocities and identify LA fibrosis (an indicator of endothelial dysfunction), but imaging remains limited in its ability to assess blood coagulation dynamics. In-silico cardiac modelling tools-such as computational fluid dynamics for blood flow, reaction-diffusion-convection equations to mimic the coagulation cascade, and surrogate flow metrics associated with endothelial damage-have grown in prevalence and advanced mechanistic understanding of thrombogenesis. However, neither technique alone can fully elucidate thrombogenicity in AF. In future, combining cardiac imaging with in-silico modelling and integrating machine learning approaches for rapid results directly from imaging data will require development under a rigorous framework of verification and clinical validation, but may pave the way towards enhanced personalised stroke risk stratification in the growing population of AF patients. This Review will focus on the significant progress in these fields.

Corrigendum: Reinforcement Learning to Improve Image-Guidance of Ablation Therapy for Atrial Fibrillation.

[This corrects the article DOI: 10.3389/fphys.2021.733139.].

Development and Characterization of Field Structured Magnetic Composites.

Polymer composites containing ferromagnetic fillers are promising for applications relating to electrical and electronic devices. In this research, the authors modified an ultraviolet light (UV) curable prepolymer to additionally cure upon heating and validated a permanent magnet-based particle alignment system toward fabricating anisotropic magnetic composites. The developed dual-cure acrylate-based resin, reinforced with ferromagnetic fillers, was first tested for its ability to polymerize through UV and heat. Then, the magnetic alignment setup was used to orient magnetic particles in the dual-cure acrylate-based resin and a heat curable epoxy resin system in a polymer casting approach. The alignment setup was subsequently integrated with a material jetting 3D printer, and the dual-cure resin was dispensed and cured in-situ using UV, followed by thermal post-curing. The resulting magnetic composites were tested for their filler loading, microstructural morphology, alignment of the easy axis of magnetization, and degree of monomer conversion. Magnetic characterization was conducted using a vibrating sample magnetometer along the in-plane and out-of-plane directions to study anisotropic properties. This research establishes a methodology to combine magnetic field induced particle alignment along with a dual-cure resin to create anisotropic magnetic composites through polymer casting and additive manufacturing.

Reinforcement Learning to Improve Image-Guidance of Ablation Therapy for Atrial Fibrillation.

Atrial fibrillation (AF) is the most common cardiac arrhythmia and currently affects more than 650,000 people in the United Kingdom alone. Catheter ablation (CA) is the only AF treatment with a long-term curative effect as it involves destroying arrhythmogenic tissue in the atria. However, its success rate is suboptimal, approximately 50% after a 2-year follow-up, and this high AF recurrence rate warrants significant improvements. Image-guidance of CA procedures have shown clinical promise, enabling the identification of key patient anatomical and pathological (such as fibrosis) features of atrial tissue, which require ablation. However, the latter approach still suffers from a lack of functional information and the need to interpret structures in the images by a clinician. Deep learning plays an increasingly important role in biomedicine, facilitating efficient diagnosis and treatment of clinical problems. This study applies deep reinforcement learning in combination with patient imaging (to provide structural information of the atria) and image-based modelling (to provide functional information) to design patient-specific CA strategies to guide clinicians and improve treatment success rates. To achieve this, patient-specific 2D left atrial (LA) models were derived from late-gadolinium enhancement (LGE) MRI scans of AF patients and were used to simulate patient-specific AF scenarios. Then a reinforcement Q-learning algorithm was created, where an ablating agent moved around the 2D LA, applying CA lesions to terminate AF and learning through feedback imposed by a reward policy. The agent achieved 84% success rate in terminating AF during training and 72% success rate in testing. Finally, AF recurrence rate was measured by attempting to re-initiate AF in the 2D atrial models after CA with 11% recurrence showing a great improvement on the existing therapies. Thus, reinforcement Q-learning algorithms can predict successful CA strategies from patient MRI data and help to improve the patient-specific guidance of CA therapy.

Tri-Planar Geometric Dimensioning and Tolerancing Characteristics of SS 316L Laser Powder Bed Fusion Process Test Artifacts and Effect of Base Plate Removal.

The precision of LPBF manufactured parts is quantified by characterizing the geometric tolerances based on the ISO 1101 standard. However, there are research gaps in the characterization of geometric tolerance of LPBF parts. A literature survey reveals three significant research gaps: (1) systematic design of benchmarks for geometric tolerance characterization with minimum experimentation; (2) holistic geometric tolerance characterization in different orientations and with varying feature sizes; and (3) a comparison of results, with and without the base plate. This research article focuses on addressing these issues by systematically designing a benchmark that can characterize geometric tolerances in three principal planar directions. The designed benchmark was simulated using the finite element method, manufactured using a commercial LPBF process using stainless steel (SS 316L) powder, and the geometric tolerances were characterized. The effect of base plate removal on the geometric tolerances was quantified. Simulation and experimental results were compared to understand tolerance variations using process variations such as base plate removal, orientation, and size. The tolerance zone variations not only validate the need for systematically designed benchmarks, but also for tri-planar characterization. Simulation and experimental result comparisons provide quantitative information about the applicability of numerical simulation for geometric tolerance prediction for the LPBF process.

Investigation of Compressive and Tensile Behavior of Stainless Steel/Dissolvable Aluminum Bimetallic Composites by Finite Element Modeling and Digital Image Correlation.

This study reports fabrication, mechanical characterization, and finite element modeling of a novel lattice structure based bimetallic composite comprising 316L stainless steel and a functional dissolvable aluminum alloy. A net-shaped 316L stainless steel lattice structure composed of diamond unit cells was fabricated by selective laser melting (SLM). The cavities in the lattice structure were then filled through vacuum-assisted melt infiltration to form the bimetallic composite. The bulk aluminum sample was also cast using the same casting parameters for comparison. The compressive and tensile behavior of 316L stainless steel lattice, bulk dissolvable aluminum, and 316L stainless steel/dissolvable aluminum bimetallic composite is studied. Comparison between experimental, finite element analysis (FEA), and digital image correlation (DIC) results are also investigated in this study. There is no notable difference in the tensile behavior of the lattice and bimetallic composite because of the weak bonding in the interface between the two constituents of the bimetallic composite, limiting load transfer from the 316L stainless steel lattice to the dissolvable aluminum matrix. However, the aluminum matrix is vital in the compressive behavior of the bimetallic composite. The dissolvable aluminum showed higher Young's modulus, yield stress, and ultimate stress than the lattice and composite in both tension and compression tests, but much less elongation. Moreover, FEA and DIC have been demonstrated to be effective and efficient methods to simulate, analyze, and verify the experimental results through juxtaposing curves on the plots and comparing strains of critical points by checking contour plots.

Toward Patient-Specific Prediction of Ablation Strategies for Atrial Fibrillation Using Deep Learning.

Atrial fibrillation (AF) is a common cardiac arrhythmia that affects 1% of the population worldwide and is associated with high levels of morbidity and mortality. Catheter ablation (CA) has become one of the first line treatments for AF, but its success rates are suboptimal, especially in the case of persistent AF. Computational approaches have shown promise in predicting the CA strategy using simulations of atrial models, as well as applying deep learning to atrial images. We propose a novel approach that combines image-based computational modelling of the atria with deep learning classifiers trained on patient-specific atrial models, which can be used to assist in CA therapy selection. Therefore, we trained a deep convolutional neural network (CNN) using a combination of (i) 122 atrial tissue images obtained by unfolding patient LGE-MRI datasets, (ii) 157 additional synthetic images derived from the patient data to enhance the training dataset, and (iii) the outcomes of 558 CA simulations to terminate several AF scenarios in the corresponding image-based atrial models. Four CNN classifiers were trained on this patient-specific dataset balanced using several techniques to predict three common CA strategies from the patient atrial images: pulmonary vein isolation (PVI), rotor-based ablation (Rotor) and fibrosis-based ablation (Fibro). The training accuracy for these classifiers ranged from 96.22 to 97.69%, while the validation accuracy was from 78.68 to 86.50%. After training, the classifiers were applied to predict CA strategies for an unseen holdout test set of atrial images, and the results were compared to outcomes of the respective image-based simulations. The highest success rate was observed in the correct prediction of the Rotor and Fibro strategies (100%), whereas the PVI class was predicted in 33.33% of the cases. In conclusion, this study provides a proof-of-concept that deep neural networks can learn from patient-specific MRI datasets and image-derived models of AF, providing a novel technology to assist in tailoring CA therapy to a patient.

Dual Graded Lattice Structures: Generation Framework and Mechanical Properties Characterization.

Additive manufacturing (AM) enables the production of complex structured parts with tailored properties. Instead of manufacturing parts as fully solid, they can be infilled with lattice structures to optimize mechanical, thermal, and other functional properties. A lattice structure is formed by the repetition of a particular unit cell based on a defined pattern. The unit cell's geometry, relative density, and size dictate the lattice structure's properties. Where certain domains of the part require denser infill compared to other domains, the functionally graded lattice structure allows for further part optimization. This manuscript consists of two main sections. In the first section, we discussed the dual graded lattice structure (DGLS) generation framework. This framework can grade both the size and the relative density or porosity of standard and custom unit cells simultaneously as a function of the structure spatial coordinates. Popular benchmark parts from different fields were used to test the framework's efficiency against different unit cell types and grading equations. In the second part, we investigated the effect of lattice structure dual grading on mechanical properties. It was found that combining both relative density and size grading fine-tunes the compressive strength, modulus of elasticity, absorbed energy, and fracture behavior of the lattice structure.

Fabrication of a Self-Healing, 3D Printable, and Reprocessable Biobased Elastomer.

A novel self-healable, fully reprocessable, and inkjet three-dimensional (3D) printable partially biobased elastomer is reported in this work. A long-chain unsaturated diacrylate monomer was first synthesized from canola oil and then cross-linked with a partially oxidized silicon-based copolymer containing free thiol groups and disulfide bonds. The elastomer is fabricated through inkjet 3D printing utilizing the photoinitiated thiol-ene click chemistry and reprocessed by compression molding exploiting the dynamic nature of disulfide bond. Self-healing is enabled by phosphine-catalyzed disulfide metathesis. The elastomer displayed a tensile strength of ∼52 kPa, a breaking strain of ∼24, and ∼86% healing efficiency at 80 °C temperature after 8 h. Moreover, the elastomer showed excellent thermal stability, and the highest thermal degradation temperature was recorded to be ∼524 °C. After reprocessing through compression molding, the elastomer fully recovered its mechanical and thermal properties. These properties of the elastomer yield an ecofriendly alternative of fossil fuel-based elastomers that can find broad applications in soft robotics, flexible wearable devices, strain sensors, health care, and next-generation energy-harvesting and -storage devices.

Rheology-Assisted Microstructure Control for Printing Magnetic Composites-Material and Process Development.

Magnetic composites play a significant role in various electrical and electronic devices. Properties of such magnetic composites depend on the particle microstructural distribution within the polymer matrix. In this study, a methodology to manufacture magnetic composites with isotropic and anisotropic particle distribution was introduced using engineered material formulations and manufacturing methods. An in-house developed material jetting 3D printer with particle alignment capability was utilized to dispense a UV curable resin formulation to the desired computer aided design (CAD) geometry. Formulations engineered using additives enabled controlling the rheological properties and the microstructure at different manufacturing process stages. Incorporating rheological additives rendered the formulation with thixotropic properties suitable for material jetting processes. Particle alignment was accomplished using a magnetic field generated using a pair of permanent magnets. Microstructure control in printed composites was observed to depend on both the developed material formulations and the manufacturing process. The rheological behavior of filler-modified polymers was characterized using rheometry, and the formulation properties were derived using mathematical models. Experimental observations were correlated with the observed mechanical behavior changes in the polymers. It was additionally observed that higher additive content controlled particle aggregation but reduced the degree of particle alignment in polymers. Directionality analysis of optical micrographs was utilized as a tool to quantify the degree of filler orientation in printed composites. Characterization of in-plane and out-of-plane magnetic properties using a superconducting quantum interference device (SQUID) magnetometer exhibited enhanced magnetic characteristics along the direction of field structuring. Results expressed in this fundamental research serve as building blocks to construct magnetic composites through material jetting-based additive manufacturing processes.

Concurrent Modelling and Experimental Investigation of Material Properties and Geometries Produced by Projection Microstereolithography.

Projection microstereolithography additive manufacturing (PµSLA-AM) systems utilize free radical photopolymerization to selectively transform liquid resins into accurate and complex, shaped, solid parts upon UV light exposure. The material properties are coupled with geometrical accuracy, implying that optimizing one response will affect the other. Material properties can be enhanced by the post-curing process, while geometry is controlled during manufacturing. This paper uses designed experiments and analytical curing models concurrently to investigate the effects of process parameters on the green material properties (after manufacturing and before applying post curing), and the geometrical accuracy of the manufactured parts. It also presents a novel accumulated energy model that considers the light absorbance of the liquid resin and solid polymer. An essential definition, named the irradiance affected zone (IAZ), is introduced to estimate the accumulated energy for each layer and to assess the feasibility of the geometries. Innovative methodologies are used to minimize the effect of irradiance irregularities on the responses and to characterize the light absorbance of liquid and cured resin. Analogous to the working curve, an empirical model is proposed to define the critical energies required to start developing the different material properties. The results of this study can be used to develop an appropriate curing scheme, to approximate an initial solution and to define constraints for projection microstereolithography geometry optimization algorithms.

National Approach to Standardize and Improve Mechanical Ventilation.

NASAM (National Approach to Standardize and Improve Mechanical Ventilation) is a national collaborative quality improvement project in Saudi Arabia. It aims to improve the care of mechanically ventilated patients by implementing evidence-based practices with the goal of reducing the rate of ventilator-associated events and therefore reducing mortality, mechanical ventilation duration and intensive care unit (ICU) length of stay. The project plans to extend the implementation to a total of 100 ICUs in collaboration with multiple health systems across the country. As of March 22, 2019, a total of 78 ICUs have registered from 6 different health sectors, 48 hospitals, and 27 cities. The leadership support in all health sectors for NASAM speaks of the commitment to improve the care of mechanically ventilated patients across the kingdom.

Efficacy Of Hydrogen Peroxide Fumigation In Improving Disinfection Of Hospital Rooms And Reducing The Number Of Microorganisms.

BACKGROUND: The hospital acquired infections are very common in any health care setting due to certain bacteria, viruses and fungi. In order to find out a solution to this problem, this preliminary study was designed to find out the efficiency of hydrogen peroxide fumigation in reducing the number of microorganisms and improving the disinfection of hospital rooms. It was a prospective cross over study, conducted in Arar Central Hospital, North region, Arar, Kingdom of Saudi Arabia, for the period of one year, from March 2015 to February 2016. Objective of the study was to determine the efficacy of hydrogen peroxide fumigation in the improvement of disinfection of hospital rooms. METHODS: A total of 10 environmental samples were taken immediately after the patient was discharged (R1), 10 after terminal cleaning (R2), and 10 after the Bioxeco hydrogen peroxide fumigation (R3) in 20 different rooms of the hospital including ICU, general medical wards and operating rooms. (T=600). RESULTS: Almost 95% rooms cultured (environmental surfaces) after patient was discharged (R1) revealed microorganism growth, 80% after terminal cleaning (R2) and 2% after Bioxeco Hydrogen Peroxide fumigation revealed growths of microorganisms like bacteria and fungi on the culture plates (R3). The highest rate of room contamination was found in the rooms where the patients had stayed for a longer period of time. CONCLUSION: Hydrogen peroxide fumigation has been proved to be an efficient disinfectant in a health care setting.

Effects of magnetic field on the spreading dynamics of an impinging ferrofluid droplet.

This paper presents a comprehensive experimental study of the effects of vertically and horizontally applied magnetic fields on the dynamics of magnetowetting and the formation of satellite droplet. Besides explaining the physics of the transient variation of different drop shape parameters, the role of a magnetic field on controlling the dynamics of spreading is also presented. Ultimately the magnetic field maneuvers the droplet spreading without altering the surface chemistry. The morphological evolution and dynamics of an impacting ferrofluid droplet has also been studied. By observing the spreading at an appropriate time scale, the contrary spreading behavior of the paramagnetic ferrofluid under the effect of magnetic field is noticed. Special attention is given to the droplet break-up and satellite droplet formation. A universal relationship is presented between the drop shape parameters before and after the impact. The destination and travel path of the satellite droplet is also analyzed in a vertical as well as horizontal magnetic field, which governs the satellite droplet merging with the already deposited parent droplet.

Intrinsically motivated reinforcement learning for human-robot interaction in the real-world.

For a natural social human-robot interaction, it is essential for a robot to learn the human-like social skills. However, learning such skills is notoriously hard due to the limited availability of direct instructions from people to teach a robot. In this paper, we propose an intrinsically motivated reinforcement learning framework in which an agent gets the intrinsic motivation-based rewards through the action-conditional predictive model. By using the proposed method, the robot learned the social skills from the human-robot interaction experiences gathered in the real uncontrolled environments. The results indicate that the robot not only acquired human-like social skills but also took more human-like decisions, on a test dataset, than a robot which received direct rewards for the task achievement.

Modeling Left Atrial Flow, Energy, Blood Heating Distribution in Response to Catheter Ablation Therapy.

Introduction: Atrial fibrillation (AF) is a widespread cardiac arrhythmia that commonly affects the left atrium (LA), causing it to quiver instead of contracting effectively. This behavior is triggered by abnormal electrical impulses at a specific site in the atrial wall. Catheter ablation (CA) treatment consists of isolating this driver site by burning the surrounding tissue to restore sinus rhythm (SR). However, evidence suggests that CA can concur to the formation of blood clots by promoting coagulation near the heat source and in regions with low flow velocity and blood stagnation. Methods: A patient-specific modeling workflow was created and applied to simulate thermal-fluid dynamics in two patients pre- and post-CA. Each model was personalized based on pre- and post-CA imaging datasets. The wall motion and anatomy were derived from SSFP Cine MRI data, while the trans-valvular flow was based on Doppler ultrasound data. The temperature distribution in the blood was modeled using a modified Pennes bioheat equation implemented in a finite-element based Navier-Stokes solver. Blood particles were also classified based on their residence time in the LA using a particle-tracking algorithm. Results: SR simulations showed multiple short-lived vortices with an average blood velocity of 0.2-0.22 m/s. In contrast, AF patients presented a slower vortex and stagnant flow in the LA appendage, with the average blood velocity reduced to 0.08-0.14 m/s. Restoration of SR also increased the blood kinetic energy and the viscous dissipation due to the presence of multiple vortices. Particle tracking showed a dramatic decrease in the percentage of blood remaining in the LA for longer than one cycle after CA (65.9 vs. 43.3% in patient A and 62.2 vs. 54.8% in patient B). Maximum temperatures of 76° and 58°C were observed when CA was performed near the appendage and in a pulmonary vein, respectively. Conclusion: This computational study presents novel models to elucidate relations between catheter temperature, patient-specific atrial anatomy and blood velocity, and predict how they change from SR to AF. The models can quantify blood flow in critical regions, including residence times and temperature distribution for different catheter positions, providing a basis for quantifying stroke risks.