Soft pneumatic actuators (SPAs) are found in mobile robots, assistive wearable devices, and rehabilitative technologies. While soft actuators have been one of the most crucial elements of technology leading the development of the soft robotics field, they fall short of force output and bandwidth requirements for many tasks. In addition, other general problems remain open, including robustness, controllability, and repeatability. The SPA-pack architecture presented here aims to satisfy these standards of reliability crucial to the field of soft robotics, while also improving the basic performance capabilities of SPAs by borrowing advantages leveraged ubiquitously in biology; namely, the structured parallel arrangement of lower power actuators to form the basis of a larger and more powerful actuator module. An SPA-pack module consisting of a number of smaller SPAs will be studied using an analytical model and physical prototype. Experimental measurements show an SPA pack to generate over 112 N linear force, while the model indicates the benefit of parallel actuator grouping over a geometrically equivalent single SPA scale as an increasing function of the number of individual actuators in the group. For a module of four actuators, a 23% increase in force production over a volumetrically equivalent single SPA is predicted and validated, while further gains appear possible up to 50%. These findings affirm the advantage of utilizing a fascicle structure for high-performance soft robotic applications over existing monolithic SPA designs. An example of high-performance soft robotic platform will be presented to demonstrate the capability of SPA-pack modules in a complete and functional system.

A temporally evolving turbulent plane jet is studied both by direct numerical simulation (DNS) and Lie symmetry analysis. The DNS is based on a high-order scheme to solve the Navier–Stokes equations for an incompressible fluid. Computations were conducted at Reynolds number $\mathit{Re}_{0}=8000$, where $\mathit{Re}_{0}$ is defined based on the initial jet thickness, $\unicode[STIX]{x1D6FF}_{0.5}(0)$, and the initial centreline velocity, $\overline{U}_{1}(0)$. A symmetry approach, known as the Lie group, is used to find symmetry transformations, and, in turn, group invariant solutions, which are also denoted as scaling laws in turbulence. This approach, which has been extensively developed to create analytical solutions of differential equations, is presently applied to the mean momentum and two-point correlation equations in a temporally evolving turbulent plane jet. The symmetry analysis of these equations allows us to derive new invariant (self-similar) solutions for the mean flow and higher moments of the velocities in the jet flow. The current DNS validates the consequence of Lie symmetry analysis and therefore confirms the establishment of novel scaling laws in turbulence. It is shown that the classical scaling law for the mean velocity is a specific form of the current scaling (which has a more general form); however, the scaling for the second and higher moments (such as Reynolds stresses) has a completely different structure compared to the classical scaling. While the failure of the classical scaling for the second moments of the fluctuating velocities has been noted from the jet data for many years, the DNS results nicely match with the present self-similar relations derived from Lie symmetry analysis. Key ingredients for the present results, in particular for the scaling laws of the higher moments, are symmetries, which are of a purely statistical nature. i.e. these symmetries are admitted by the moment equations, however, they are not observed by the original Navier–Stokes equations.

No abstract

No abstract

Across jurisdictions, government and health insurance providers hold a large amount of data from patient interactions with the healthcare system. We aimed to develop a machine learning-based model for predicting adverse outcomes due to diabetes complications using administrative health data from the single-payer health system in Ontario, Canada. A Gradient Boosting Decision Tree model was trained on data from 1,029,366 patients, validated on 272,864 patients, and tested on 265,406 patients. Discrimination was assessed using the AUC statistic and calibration was assessed visually using calibration plots overall and across population subgroups. Our model predicting three-year risk of adverse outcomes due to diabetes complications (hyper/hypoglycemia, tissue infection, retinopathy, cardiovascular events, amputation) included 700 features from multiple diverse data sources and had strong discrimination (average test AUC = 77.7, range 77.7–77.9). Through the design and validation of a high-performance model to predict diabetes complications adverse outcomes at the population level, we demonstrate the potential of machine learning and administrative health data to inform health planning and healthcare resource allocation for diabetes management.

No abstract

Key Points Question Can a machine learning model trained on routinely collected administrative health data be used to accurately predict the onset of type 2 diabetes at the population level? Findings In this decision analytical model study of 2.1 million residents in Ontario, Canada, a machine learning model was developed with high discrimination, population-level calibration, and calibration across population subgroups. Meaning Study results suggest that machine learning and administrative health data can be used to create population health planning tools that accurately discriminate between high- and low-risk groups to guide investments and targeted interventions for diabetes prevention.

Inspired by the success of self attention mechanism and Transformer architecture in sequence transduction and image generation applications, we propose novel self attention-based architectures to improve the performance of adversarial latent codebased schemes in text generation. Adversarial latent code-based text generation has recently gained a lot of attention due to its promising results. In this paper, we take a step to fortify the architectures used in these setups, specifically AAE and ARAE. We benchmark two latent code-based methods (AAE and ARAE) designed based on adversarial setups. In our experiments, the Google sentence compression dataset is utilized to compare our method with these methods using various objective and subjective measures. The experiments demonstrate the proposed (self) attention-based models outperform the state-of-the-art in adversarial code-based text generation. *