An ultra energy-efficient hardware platform for neuromorphic computing enabled by 2D-TMD tunnel-FETs.

Brain-like energy-efficient computing has remained elusive for neuromorphic (NM) circuits and hardware platform implementations despite decades of research. In this work we reveal the opportunity to significantly improve the energy efficiency of digital neuromorphic hardware by introducing NM circuits employing two-dimensional (2D) transition metal dichalcogenide (TMD) layered channel material-based tunnel-field-effect transistors (TFETs). Our novel leaky-integrate-fire (LIF) based digital NM circuit along with its Hebbian learning circuitry operates at a wide range of supply voltages, frequencies, and activity factors, enabling two orders of magnitude higher energy-efficient computing that is difficult to achieve with conventional material and/or device platforms, specifically the silicon-based 7 nm low-standby-power FinFET technology. Our innovative 2D-TFET based NM circuit paves the way toward brain-like energy-efficient computing that can unleash major transformations in future AI and data analytics platforms.

Urban-rural disparity in global estimation of PM2·5 household air pollution and its attributable health burden.

BACKGROUND: Polluting fuels and inefficient stove technologies are still a leading cause of premature deaths worldwide, particularly in low-income and middle-income countries. Previous studies of global household air pollution (HAP) have neither considered the estimation of PM2·5 at national level nor the corresponding attributable mortality burden. Additionally, the effects of climate and ambient air pollution on the global estimation of HAP-PM2·5 exposure for different urban and rural settings remain largely unknown. In this study, we include climatic effects to estimate the HAP-PM2·5 exposure from different fuel types and stove technologies in rural and urban settings separately and the related attributable global mortality burden. METHODS: Bayesian hierarchical models were developed to estimate an annual average HAP-PM2·5 personal exposure and HAP-PM2·5 indoor concentration (including both outdoor and indoor sources). Model variables were selected from sample data in 282 peer-reviewed studies drawn and updated from the WHO Global HAP dataset. The PM2·5 exposure coefficients from the developed model were applied to the external datasets to predict the HAP-PM2·5 exposure globally (personal exposure in 62 countries and indoor concentration in 69 countries). Attributable mortality rate was estimated using a comparative risk assessment approach. Using weighted averages, the national level 24 h average HAP-PM2·5 exposure due to polluting and clean fuels and related death rate per 100 000 population were estimated. FINDINGS: In 2020, household use of polluting solid fuels for cooking and heating led to a national-level average personal exposure of 151 µg/m3 (95% CI 133-169), with rural households having an average of 171 µg/m3 (153-189) and urban households an average of 92 µg/m3 (77-106). Use of clean fuels gave rise to a national-level average personal exposure of 69 µg/m3 (62-76), with a rural average of 76 µg/m3 (69-83) and an urban average of 49 µg/m3 (46-53). Personal exposure-attributable premature mortality (per 100 000 population) from the use of polluting solid fuels at national level was on average 78 (95% CI 69-87), with a rural average of 82 (73-90) and an urban average of 66 (57-75). The average attributable premature mortality (per 100 000 population) from the use of clean fuels at the national level is 62 (54-70), with a rural average of 66 (58-74) and an urban average of 52 (47-57). The estimated HAP-PM2·5 indoor concentration shows that the use of polluting solid fuels resulted in a national-level average of 412 µg/m3 (95% CI 353-471), with a rural average of 514 µg/m3 (446-582) and an urban average of 149 µg/m3 (126-173). The use of clean fuels (gas and electricity) led to an average PM2·5 indoor concentration of 135 µg/m3 (117-153), with a rural average of 174 µg/m3 (154-195) and an urban average of 71 µg/m3 (63-80). Using time-weighted HAP-PM2·5 indoor concentrations, the attributable premature death rate (per 100 000 population) from the use of polluting solid fuels at the national level is on average 78 (95% CI 72-84), the rural average being 84 (78-91) and the urban average 60 (54-66). From the use of clean fuels, the average attributable premature death rate (per 100 000 population) at the national level is 59 (53-64), the rural average being 68 (62-74) and the urban average 45 (41-50). INTERPRETATION: A shift from polluting to clean fuels can reduce the average PM2·5 personal exposure by 53% and thereby lower the death rate. For all fuel types, the estimated average HAP-PM2·5 personal exposure and indoor concentrations exceed the WHO's Interim Target-1 average annual threshold. Policy interventions are urgently needed to greatly increase the use of clean fuels and stove technologies by 2030 to achieve the goal of affordable clean energy access, as set by the UN in 2015, and address health inequities in urban-rural settings. FUNDING: Wellcome Trust, The Lancet Countdown, the Engineering and Physical Sciences Research Council, and the Natural Environment Research Council.

Do verbal coaching cues and analogies affect motor skill performance in youth populations?

PURPOSE: The way coaching cues are worded can impact on the quality with which a subsequent motor skill is executed. However, there have been few investigations on the effect of coaching cues on basic motor skill performance in youths. METHOD: Across several international locations, a series of experiments were undertaken to determine the effect of external coaching cues (EC), internal coaching cues (IC), analogies with a directional component (ADC) and neutral control cues on sprint time (20 m) and vertical jump height in youth performers. These data were combined using internal meta-analytical techniques to pool results across each test location. This approach was amalgamated with a repeated-measures analysis to determine if there were any differences between the ECs, ICs and ADCs within the different experiments. RESULTS: 173 participants took part. There were no differences between the neutral control and experimental cues in any of the internal meta-analyses except where the control was superior to the IC for vertical jump (d = -0.30, [-0.54, -0.05], p = 0.02). Just three of eleven repeated-measures analyses showed significant differences between the cues at each experimental location. Where significant differences were noted, the control cue was most effective with some limited evidence supporting the use of ADCs (d = 0.32 to 0.62). CONCLUSION: These results suggest the type of cue or analogy provided to a youth performer has little subsequent effect on sprint or jump performance. Accordingly, coaches might take a more specific approach that is suited to the level or preferences of a particular individual.

Temperature and strain sensitivities of a groove bonded fiber Bragg grating at room and cryogenic temperatures.

Optical fiber Bragg gratings (FBGs) are well suited for applications as temperature or/and strain sensors in harsh environments, e.g., detecting thermal hot spots in high-temperature superconductor (HTS) fusion energy magnets at cryogenic temperatures and high radiation environments. To maximize the signal-to-noise ratio (SNR) of the FBGs to a hot spot, we propose to have them mounted in V-shaped grooves of HTS' copper former. To investigate the differences between different adhesives on transferring strain and heat in this configuration, five arrays of FBGs are mounted in the V-shaped grooves of a copper dog-bone using Scotch-Weld epoxy, Stycast 2850 FT, Apiezon N, and Loctite 5145 silicone. The copper is cycled through tensile forces in a modified universal tensile tester, subjected to a thermal cycle between 293 K and 77 K, and exposed to heat pulse propagations at 293 K and 80 K. The FBGs that are bonded using Stycast show the highest temperature and strain sensitivities at room and cryogenic temperatures. No major differences in the temperature and strain sensitivities have been found between Ormocer and polyimide coated FBGs. Apiezon N is found to transfer strain consistently well below 245 K, which is comparable with other bonding materials in the temperature range between 77 K and 110 K. The FBGs bonded with the four adhesives in the V-groove configurations are shown to have comparable SNRs to a temperature rise of 20 K at 80 K. This paper emphasizes the importance of maximizing the thermal strain transferred from the host material through the bonding agents to achieve high temperature sensitivity of FBGs.

Safety and effectiveness of ustekinumab in elderly Crohn's disease patients.

OBJECTIVE: Anti-tumour necrosis factor (TNF) agents are associated with increased infection risk among elderly IBD patients, but little is known about non anti-TNF biologics in this cohort. We examined the safety and effectiveness of ustekinumab in elderly Crohn's patients. METHODS: This retrospective multi-centre cohort study included Crohn's patients ≥60-years old who commenced ustekinumab. We recorded Harvey-Bradshaw index (HBI), concomitant steroid therapy, treatment persistence and new infections or malignancies. Primary outcome was serious infections requiring hospitalisation. RESULTS: Seventy patients were included, with median age of 68 years. 43 (61.4%) had prior anti-TNF exposure, and 15 (21.4%) vedolizumab. Median treatment duration was 12 months, totalling 84 patient-years. Nine serious infections were reported, incidence 106.7/1000 patient-years. Systemic steroids were associated with increased risk of serious infections [odds ratio (OR) 7.83, 95% confidence interval (CI): 1.44-44.32, P = 0.02]. There were 27 "non-serious" infections; 321.4/1000 patient-years. Charlson co-morbidity index (OR 1.49, 95% CI: 1.05-2.12, P = 0.03) and steroid exposure (OR 44.10, 95% CI: 1.75-1112.10, P = 0.02) increased non-serious infection risk (P < 0.05). Mean HBI improved from 8.13 to 4.64 at 6 months and 4.10 at last follow up (P < 0.0001). 12-month treatment persistence was 55.7% (N = 39); 34 (48.6%) were steroid-free. CONCLUSION: Ustekinumab was safe and effective in a cohort of elderly Crohn's disease patients. Infections were mostly mild, not resulting in therapy discontinuation. Serious infection risk was comparable to previously reported rates with anti-TNF agents. Steroid exposure was associated with an increased serious infection risk.

Vector Symbolic Architectures as a Computing Framework for Emerging Hardware.

This article reviews recent progress in the development of the computing framework Vector Symbolic Architectures (also known as Hyperdimensional Computing). This framework is well suited for implementation in stochastic, emerging hardware and it naturally expresses the types of cognitive operations required for Artificial Intelligence (AI). We demonstrate in this article that the field-like algebraic structure of Vector Symbolic Architectures offers simple but powerful operations on high-dimensional vectors that can support all data structures and manipulations relevant to modern computing. In addition, we illustrate the distinguishing feature of Vector Symbolic Architectures, "computing in superposition," which sets it apart from conventional computing. It also opens the door to efficient solutions to the difficult combinatorial search problems inherent in AI applications. We sketch ways of demonstrating that Vector Symbolic Architectures are computationally universal. We see them acting as a framework for computing with distributed representations that can play a role of an abstraction layer for emerging computing hardware. This article serves as a reference for computer architects by illustrating the philosophy behind Vector Symbolic Architectures, techniques of distributed computing with them, and their relevance to emerging computing hardware, such as neuromorphic computing.

Dual Convolutional Neural Networks for Breast Mass Segmentation and Diagnosis in Mammography.

Deep convolutional neural networks (CNNs) have emerged as a new paradigm for Mammogram diagnosis. Contemporary CNN-based computer-aided-diagnosis systems (CADs) for breast cancer directly extract latent features from input mammogram image and ignore the importance of morphological features. In this paper, we introduce a novel end-to-end deep learning framework for mammogram image processing, which computes mass segmentation and simultaneously predicts diagnosis results. Specifically, our method is constructed in a dual-path architecture that solves the mapping in a dual-problem manner, with an additional consideration of important shape and boundary knowledge. One path, called the Locality Preserving Learner (LPL), is devoted to hierarchically extracting and exploiting intrinsic features of the input. Whereas the other path, called the Conditional Graph Learner (CGL), focuses on generating geometrical features via modeling pixel-wise image to mask correlations. By integrating the two learners, both the cancer semantics and cancer representations are well learned, and the component learning paths in return complement each other, contributing an improvement to the mass segmentation and cancer classification problem at the same time. In addition, by integrating an automatic detection set-up, the DualCoreNet achieves fully automatic breast cancer diagnosis practically. Experimental results show that in benchmark DDSM dataset, DualCoreNet has outperformed other related works in both segmentation and classification tasks, achieving 92.27% DI coefficient and 0.85 AUC score. In another benchmark INbreast dataset, DualCoreNet achieves the best mammography segmentation (93.69% DI coefficient) and competitive classification performance (0.93 AUC score).

The wicked problem of waste management: An attention-based analysis of stakeholder behaviours.

Surging amounts of waste are reported globally and especially in lower-income countries, with negative consequences for health and the environment. Increasing concern has been raised for the limited progress achieved in practice by diverse sets of policies and programmes. Waste management is a wicked problem characterised by multilayered interdependencies, complex social dynamics and webs of stakeholders. Interactions among these generate unpredictable outcomes that can be missed by decision makers through their understanding and framing of their context. This article aims to identify possible sources of persistent problems by focussing on what captures, shapes and limits the attention of stakeholders and decision-makers, drawing on the attention-based view from organisation theory. The theory describes the process through which issues and opportunities are noticed and how these are translated into actions, by focussing on the influencers at the individual, organisational and context scale. Views on issues and opportunities for waste management were collected in a series of fieldwork activities from 60 participants representing seven main types of stakeholders in the typical lower-middle income Kenyan city of Kisumu. Through a thematic analysis guided by the attention-based view, we identified patterns and misalignment of views, especially between government, community-based organisations and residents, which may contribute to persistent waste problems in Kisumu. Some point to detrimental waste handling practices, from separation to collection and treatment, as the main cause of issues. For others, these practices are due to a poor control of such practices and enforcement of the law. This study's major theoretical contribution is extending the application of attention theory to multi-stakeholder problems and to non-formalized organisations, namely residents and to the new field of waste management. This novel lens contributes a greater understanding of waste issues and their management in Africa that is relevant to policy and future research. By revealing the "wickedness" of the waste problem, we point to the need for a holistic and systems-based policy approach to limit further unintended consequences.

A multi-scalar perspective on health and urban housing: an umbrella review.

With more than half the world's population living in cities, understanding how the built environment impacts human health at different urban scales is crucial. To be able to shape cities for health, an understanding is needed of planetary health impacts, which encompass the human health impacts of human-caused disruptions on the Earth's natural ecosystems. This umbrella review maps health evidence across the spatial scales of the built environment (building; neighbourhood; and wider system, including city, regional and planetary levels), with a specific focus on urban housing. Systematic reviews published in English between January 2011 and December 2020 were searched across 20 databases, with 1176 articles identified and 124 articles screened for inclusion. Findings suggests that most evidence reports on health determinants at the neighbourhood level, such as greenspace, physical and socio-economic conditions, transport infrastructure and access to local services. Physical health outcomes are also primarily reported, with an emerging interest in mental health outcomes. There is little evidence on planetary health outcomes and significant gaps in the research literature are identified. Based on these findings, three potential directions are identified for future research.

Projecting the impacts of housing on temperature-related mortality in London during typical future years.

Climate change means the UK will experience warmer winters and hotter summers in the future. Concurrent energy efficiency improvements to housing may modify indoor exposures to heat or cold, while population aging may increase susceptibility to temperature-related mortality. We estimate heat and cold mortality and energy consumption in London for typical (non-extreme) future climates, given projected changes in population and housing. Building physics models are used to simulate summertime and wintertime indoor temperatures and space heating energy consumption of London dwellings for 'baseline' (2005-2014) and future (2030s, 2050s) periods using data from the English Housing Survey, historical weather data, and projected future weather data with temperatures representative of 'typical' years. Linking to population projections, we calculate future heat and cold attributable mortality and energy consumption with demolition, construction, and alternative scenarios of energy efficiency retrofit. At current retrofit rates, around 168-174 annual cold-related deaths per million population would typically be avoided by the 2050s, or 261-269 deaths per million under ambitious retrofit rates. Annual heat deaths would typically increase by 1 per million per year under the current retrofit rate, and 12-13 per million under ambitious rates without population adaptation to heat. During typical future summers, an estimated 38-73% of heat-related deaths can be avoided using external shutters on windows, with their effectiveness lower during hotter weather. Despite warmer winters, ambitious retrofit rates are necessary to reduce typical annual energy consumption for heating below baseline levels, assuming no improvement in heating system efficiencies. Concerns over future overheating in energy efficient housing are valid but increases in heat attributable mortality during typical and hot (but not extreme) summers are more than offset by significant reductions in cold mortality and easily mitigated using passive measures. More ambitious retrofit rates are critical to reduce energy consumption and offer co-benefits for reducing cold-related mortality.

Bayesian Activity Estimation and Uncertainty Quantification of Spent Nuclear Fuel Using Passive Gamma Emission Tomography.

In this paper, we address the problem of activity estimation in passive gamma emission tomography (PGET) of spent nuclear fuel. Two different noise models are considered and compared, namely, the isotropic Gaussian and the Poisson noise models. The problem is formulated within a Bayesian framework as a linear inverse problem and prior distributions are assigned to the unknown model parameters. In particular, a Bernoulli-truncated Gaussian prior model is considered to promote sparse pin configurations. A Markov chain Monte Carlo (MCMC) method, based on a split and augmented Gibbs sampler, is then used to sample the posterior distribution of the unknown parameters. The proposed algorithm is first validated by simulations conducted using synthetic data, generated using the nominal models. We then consider more realistic data simulated using a bespoke simulator, whose forward model is non-linear and not available analytically. In that case, the linear models used are mis-specified and we analyse their robustness for activity estimation. The results demonstrate superior performance of the proposed approach in estimating the pin activities in different assembly patterns, in addition to being able to quantify their uncertainty measures, in comparison with existing methods.

Climate change projections for sustainable and healthy cities.

The ambition to develop sustainable and healthy cities requires city-specific policy and practice founded on a multidisciplinary evidence base, including projections of human-induced climate change. A cascade of climate models of increasing complexity and resolution is reviewed, which provides the basis for constructing climate projections-from global climate models with a typical horizontal resolution of a few hundred kilometres, through regional climate models at 12-50 km to convection-permitting models at 1 km resolution that permit the representation of urban induced climates. Different approaches to modelling the urban heat island (UHI) are also reviewed-focusing on how climate model outputs can be adjusted and coupled with urban canopy models to better represent UHI intensity, its impacts and variability. The latter can be due to changes induced by urbanisation or to climate change itself. City interventions such as greater use of green infrastructure also have an effect on the UHI and can help to reduce adverse health impacts such as heat stress and the mortality associated with increasing heat. Examples for the Complex Urban Systems for Sustainability and Health (CUSSH) partner cities of London, Rennes, Kisumu, Nairobi, Beijing and Ningbo illustrate how cities could potentially make use of more detailed models and projections to develop and evaluate policies and practices targeted at their specific environmental and health priorities. PRACTICE RELEVANCE: Large-scale climate projections for the coming decades show robust trends in rising air temperatures, including more warm days and nights, and longer/more intense warm spells and heatwaves. This paper describes how more complex and higher resolution regional climate and urban canopy models can be combined with the aim of better understanding and quantifying how these larger scale patterns of change may be modified at the city or finer scale. These modifications may arise due to urbanisation and effects such as the UHI, as well as city interventions such as the greater use of grey and green infrastructures.There is potential danger in generalising from one city to another-under certain conditions some cities may experience an urban cool island, or little future intensification of the UHI, for example. City-specific, tailored climate projections combined with tailored health impact models contribute to an evidence base that supports built environment professionals, urban planners and policymakers to ensure designs for buildings and urban areas are fit for future climates.

Systemic inequalities in indoor air pollution exposure in London, UK.

Deprived communities in many cities are exposed to higher levels of outdoor air pollution, and there is increasing evidence of similar disparities for indoor air pollution exposure. There is a need to understand the drivers for this exposure disparity in order to develop effective interventions aimed at improving population health and reducing health inequities. With a focus on London, UK, this paper assembles evidence to examine why indoor exposure to PM2.5, NOx and CO may disproportionately impact low-income groups. In particular, five factors are explored, namely: housing location and ambient outdoor levels of pollution; housing characteristics, including ventilation properties and internal sources of pollution; occupant behaviours; time spent indoors; and underlying health conditions. Evidence is drawn from various sources, including building physics models, modelled outdoor air pollution levels, time-activity surveys, housing stock surveys, geographical data, and peer-reviewed research. A systems framework is then proposed to integrate these factors, highlighting how exposure to high levels of indoor air pollution in low-income homes is in large part due to factors beyond the control of occupants, and is therefore an area of systemic inequality. POLICY RELEVANCE: There is increasing public and political awareness of the impact of air pollution on public health. Strong scientific evidence links exposure to air pollution with morbidity and mortality. Deprived communities may be more affected, however, with limited evidence on how deprivation may influence their personal exposure to air pollution, both outdoors and indoors. This paper describes different factors that may lead to low-income households being exposed to higher levels of indoor air pollution than the general population, using available data and models for London (i.e. living in areas of higher outdoor air pollution, in poor-quality housing, undertaking more pollution-generating activities indoors and spending more time indoors). A systems approach is used to show how these factors lead to systemic exposure inequalities, with low-income households having limited opportunities to improve their indoor air quality. This paper can inform actions and public policies to reduce environmental health inequalities, considering both indoor and outdoor air.

Compressive MRI quantification using convex spatiotemporal priors and deep encoder-decoder networks.

We propose a dictionary-matching-free pipeline for multi-parametric quantitative MRI image computing. Our approach has two stages based on compressed sensing reconstruction and deep learned quantitative inference. The reconstruction phase is convex and incorporates efficient spatiotemporal regularisations within an accelerated iterative shrinkage algorithm. This minimises the under-sampling (aliasing) artefacts from aggressively short scan times. The learned quantitative inference phase is purely trained on physical simulations (Bloch equations) that are flexible for producing rich training samples. We propose a deep and compact encoder-decoder network with residual blocks in order to embed Bloch manifold projections through multi-scale piecewise affine approximations, and to replace the non-scalable dictionary-matching baseline. Tested on a number of datasets we demonstrate effectiveness of the proposed scheme for recovering accurate and consistent quantitative information from novel and aggressively subsampled 2D/3D quantitative MRI acquisition protocols.

Collaborative working in health and social care: Lessons learned from post-hoc preliminary findings of a young families' pregnancy to age 2 project in South Wales, United Kingdom.

Children of young and socially disadvantaged parents are more likely to experience adverse outcomes. In response to this, a unique young families' project in Swansea, UK, was created, which drew together a team of multi-agency professionals, to support people aged 16-24 from 17 weeks of pregnancy throughout 1,001 days of the child's life. The aim of the JIGSO (the Welsh word for Jigsaw) project is for young people to reach their potential as parents and to break the cycle of health and social inequality. This evaluation analysed routinely collected data held by the project from January 2017 to December 2018 exploring health and social outcomes, including smoking and alcohol use in pregnancy, breastfeeding, maternal diet and social services outcomes. Outcomes were compared to local and national averages, where available. Data relating to parenting knowledge and skills were available via records of 10-point Likert scales, one collected at the start of the JIGSO involvement and one around 4-6 months later. Findings showed higher than average levels of breastfeeding initiation and lower smoking and alcohol use in pregnancy. Parents also reported enhanced knowledge and confidence in their child care skills, as well as improved family relationships. Parents with high levels of engagement with JIGSO also appeared to have positive outcomes with Social Services (their child's name was removed from child protection register or their case was closed to social services). This was a post-hoc evaluation, not an intervention study or trial, and thus findings must be interpreted with caution. Despite this, the findings are promising and more prospective research exploring similar services is required.