Predicting Urban Traffic Under Extreme Weather by Deep Learning Method with Disaster Knowledge

Video-based human action recognition is one of the most important and challenging areas of research in the field of computer vision. Human action recognition has found many pragmatic applications in video surveillance, human-computer interaction, entertainment, autonomous driving, etc. Owing to the recent development of deep learning methods for human action recognition, the performance of action recognition has significantly enhanced for challenging datasets. Deep learning techniques are mainly used for recognizing actions in images and videos comprising of Euclidean data. A recent development in deep learning methods is the extension of these techniques to non-Euclidean data or graph data with many nodes and edges. Human body skeleton resembles a graph, therefore, the graph convolutional network (GCN) is applicable to the non-Euclidean body skeleton. In the past few years, GCN has emerged as an important tool for skeleton-based action recognition. Therefore, we conduct a survey using GCN methods for action recognition. Herein, we present a comprehensive overview of recent GCN techniques for action recognition, propose a taxonomy for the categorization of GCN techniques for action recognition, carry out a detailed study of the benchmark datasets, enlist relevant resources and open-source codes, and finally provide an outline for future research directions and trends. To the best of authors' knowledge, this is the first survey for action recognition using GCN techniques.

The increasing intensity and frequency of extreme weather events resulting from climate change have led to grid outages and other negative consequences. To ensure the resilience of buildings which serve as primary shelters for occupants, resilient strategies are being developed to improve their ability to withstand these extreme events (e.g., building upgrades and renewable energy generators and storage). However, a crucial step towards creating a resilient built environment is accurately estimating building performance during such conditions using historical extreme climate change-induced weather events. To conduct Building Performance Simulation (BPS) in extreme conditions, such as weather events induced by climate change, it is essential to utilize Actual Meteorological Year (AMY) weather files instead of Typical Meteorological Year (TMY) files. AMY files capture the precise climatic conditions during extreme weather events, enabling accurate simulation of such scenarios. These weather files provide valuable data that can be used to assess the vulnerabilities and resilience of buildings to extreme weather events. By analyzing past events and their impacts using BPS tools, we can gain insights into the specific weaknesses and areas that require improvement. This approach applies to both existing buildings needing climate change-resilient retrofits and new building designs that must be compatible with future climatic conditions. Moreover, the intensification and frequency increase of these extreme weather events makes developing adaptation and resilient-building measures imperative. This involves understanding the potential losses that households may experience due to the intensification of extreme events and developing farsighted coping strategies and climate-proof resilient-building initiatives. However, addressing the knowledge gap caused by the absence of an AMY weather file dataset of extreme events is essential. This will allow for accurate BPS during past extreme climate change-induced weather events. To fill this gap, this article introduces a comprehensive .epw format weather file dataset focusing on historical extreme weather events in Canada. This collection encompasses a diverse array of past extreme climate change occurrences in various locations, with potential for future expansion to include additional locations and countries. This dataset enables energy simulations for different types of buildings and considers a diverse range of historical weather conditions, allowing for better estimation of thermal performance.

Objectives: It is clinically important to evaluate the performance of a newly developed blood pressure (BP) measurement method under different measurement conditions. This study aims to evaluate the performance of using deep learning-based method to measure BPs and BP change under non-resting conditions.Materials and methods: Forty healthy subjects were studied. Systolic and diastolic BPs (SBPs and DBPs) were measured under four conditions using deep learning and manual auscultatory method. The agreement between BPs determined by the two methods were analysed under different conditions. The performance of using deep learning-based method to measure BP changes was finally evaluated.Results: There were no significant BPs differences between two methods under all measurement conditions (all p > .1). SBP and DBP measured by deep learning method changed significantly in comparison with the resting condition: decreased by 2.3 and 4.2 mmHg with deeper breathing (both p < .05), increased by 3.6 and 6.4 mmHg with talking, and increased by 5.9 and 5.8 mmHg with arm movement (all p < .05). There were no significant differences in BP changes measured by two methods (all p > .4, except for SBP change with deeper breathing).Conclusion: This study demonstrated that the deep learning method could achieve accurate BP measurement under both resting and non-resting conditions.Key messagesAccurate and reliable blood pressure measurement is clinically important. We evaluated the performance of our developed deep learning-based blood pressure measurement method under resting and non-resting measurement conditions.The deep learning-based method could achieve accurate BP measurement under both resting and non-resting measurement conditions.

Automatic modulation classification (AMC) of underwater acoustic communication signals is of great significance in national defense and marine military. Accurate modulation classification methods can make great contributions to accurately grasping the parameters and characteristics of enemy communication systems. While a poor underwater acoustic channel makes it difficult to classify the modulation types correctly. Feature extraction and deep learning methods have proven to be effective methods for the modulation classification of underwater acoustic communication signals, but their performance is still limited by the complex underwater communication environment. Graph convolution networks (GCN) can learn the graph structured information of the data, making it an effective method for processing structured data. To improve the stability and robustness of AMC in underwater channels, we combined the feature extraction and deep learning methods by fusing the multi-domain features and deep features using GCN. The proposed method takes the relationships among the different multi-domain features and deep features into account. Firstly, a feature graph was built using the properties of the features. Secondly, multi-domain features were extracted from the received signals and deep features were extracted from the signals using a deep neural network. Thirdly, we constructed the input of GCN using these features and the graph. Then, the multi-domain features and deep features were fused by the GCN. Finally, we classified the modulation types using the output of GCN by way of a softmax layer. We conducted the experiments on a simulated dataset and a real-world dataset, respectively. The results show that the AMC based on GCN can achieve a significant improvement in performance compared to the current state-of-the-art methods. Our approach is robust in underwater acoustic channels.

People experiencing poverty and inequality will be affected first and worst by the impacts of climate change to infrastructure and human settlements, including those caused by increasingly frequent and intense extreme weather events and natural disasters. They have the least capacity to cope, to adapt, to move and to recover. Community service organisations (CSOs) play a critical role in supporting individuals, families and communities experiencing poverty and inequality to build resilience and respond to adverse changes in circumstances. As such, the services they provide comprise a critical component of social infrastructure in human settlements. However, very little is understood about CSOs own vulnerability to – or their role in managing and mitigating risks to their clients and the community from – climate change impacts to physical infrastructure. The Extreme Weather, Climate Change and the Community Sector – Risks and Adaptations project examined the relationship between physical and social infrastructure (in the form of CSO service provision). Specifically, the ways in which the climate-driven failure of CSO service delivery worsens risks to the individuals and communities they serve and, on the other hand, how preparedness may reduce vulnerability to climate change and extreme weather impacts to human settlements and infrastructure. The research comprised a comprehensive and critical scoping, examination and review of existing research findings and an audit, examination and judgment-based evaluation of the current vulnerabilities and capacities of CSOs under projected climate change scenarios. It employed three key methods of consultation and data collection. A literature review examined research conducted to date in Australia and comparative countries internationally on the vulnerability and climate change adaptation needs of CSOs. A program of 10 Community Sector Professional Climate Workshops consulted over 150 CSO representatives to develop a qualitative record of extreme event and climate change risks and corresponding adaptation strategies specific to CSOs. A national survey of CSOs, which resulted in the participation of approximately 500 organisations, produced a quantitative data set about the nature of CSO vulnerability to climate change and extreme weather impacts to infrastructure, whether and how CSOs are approaching the adaptation task and key barriers to adaptation. While the methods employed and the absence of empirical data sets quantifying CSO vulnerability to climate change impacts create limitations to the evidence-base produced, findings from the research suggest that CSOs are highly vulnerable and not well prepared to respond to climate change and extreme weather impacts to physical infrastructure and that this underlying organisational vulnerability worsens the vulnerability of people experiencing poverty and inequality to climate change. However, the project results indicate that if well adapted, CSOs have the willingness, specialist skills, assets and capacity to make a major contribution to the resilience and adaptive capacity of their clients and the community more broadly (sections of which will be plunged into adversity by extreme events). Despite this willingness, the evidence presented shows that few CSOs have undertaken significant action to prepare for climate change and worsening extreme weather events. Key barriers to adaptation identified through the research are inadequate financial resources, lack of institutionalised knowledge and skills for adaptation and the belief that climate change adaptation is beyond the scope of CSOs core business. On the other hand, key indicators of organisational resilience to climate change and extreme weather impacts include: level of knowledge about extreme weather risks, past experience of an extreme weather event and organisational size. Given its size, scope and the critical role the Australian community sector plays in building client and community resilience and in assisting communities to respond to and recover from the devastating impacts of extreme weather events and natural disasters, the research identifies serious gaps in both the policy frameworks and the research base required to ensure the sector’s resilience and adaptive capacity – gaps which appear to have already had serious consequences. To address these gaps, a series of recommendations has been prepared to enable the development and implementation of a comprehensive, sector-specific adaptation and preparedness program, which includes mechanisms to institutionalise knowledge and skills, streamlined tools appropriate to the needs and capacity of a diverse range of organisations and a benchmarking system to allow progress towards resilience and preparedness to be monitored. Future research priorities for adaptation in this sector have also been identified.

This paper reports the development and results of a model exploring the resilience of the Dutch electricity transmission infrastructure to extreme weather events. Climate change is anticipated to result in an increase in the frequency and severity of extreme weather events over the coming decades. Situated in a low-lying coastal delta, the Netherlands may be particularly exposed to certain types of extreme weather(-induced) events. The degree to which the country’s electricity network may prove resilient in the face of these future events is an open question. The model focuses on two types of extreme events – floods and heat waves – and assesses two types of adaptation measures – substation flood protections and demand-side management. The model employs a network-based approach in assessing infrastructure resilience – explicitly representing the structure and properties of the Dutch transmission infrastructure – and extends previous work by accounting for key power system characteristics such as capacity constraints and cascading failures. From a practice perspective, the results offer a first indication of the vulnerability of the Dutch electricity transmission infrastructure in the context of climate change. These results suggest that the network displays some vulnerability to both floods and heat waves. Both types of adaptation measures tested are found to enhance resilience, though substation flood protection shows greater benefits. Whilst the model was specifically developed for the study of electricity networks, we anticipate that this method may also be applicable to other types of transport infrastructures.

This paper analyzes the change characteristics of extreme weather and climate events,such as heavy rain,drought,high temperature ≥35 ℃,low temperature ≤-20 ℃,frost,gale,sandstorm,thunderstorm and hail over the Hexi Corridor using the data of annual occurring days of extreme weather and climate events from 18 meteorological stations in the corridor during the period of 1958-2006 and the statistical methods,such as polynomial,linear trend,Mann-Kendall test,Moving T-test and Morlet wavelet analysis.The results show that the annual days of occurring some extreme weather events,such as gale,sandstorm,low temperature ≤-20 ℃,frost,thunderstorm and hail,over the Hexi Corridor in recent 49 years have been significantly decreased,but the annual days of occurring high temperature ≥35 ℃ have been obviously increased.Of which the abrupt change of occurrence of gale,sandstorm,high temperature ≥35 ℃ and frost occurred around 1996,1989,1996 and 1998 respectively.Generally,the change of the frequencies of heavy rain and drought is not so obvious.In the early 1990s,the annual days of occurring low temperature ≤-20 ℃ and frost were decreased,but those of high temperature ≥35 ℃,heavy rain and drought were increased,and these trends met the abrupt warming in the Hexi Corridor.There is a long-term periodical change in high temperature and drought but a staggered one in sandstorm and thunderstorm,and the periodical change of other extreme weather and climate events is not so obvious.

Instrumental weather records (1880–2020s) from eastern North America were analyzed to characterize the regional patterns and drivers of seasonal extreme weather (snow, rain, high and low temperatures). Using agglomerative hierarchical clustering of extreme weather data, the region was divided into three subregions that are influenced by coastal‐marine gradients and latitudinal factors. Subsequent analyses were performed on high‐quality stations from each subregion and results compared between one another. Long‐term locally weighted linear regressions delineated long‐term changes in extreme weather, and a combination of spectral analysis, continuous wavelet transforms, and cross wavelet transforms were used to identify periodic components in the data. Regional extreme weather is generally periodic, composed of interannual to interdecadal‐scale oscillations and driven by several natural climatic oscillations. The most important such oscillation is the 11‐year Schwabe Solar Cycle, which has a strong and continuous effect on regional extreme weather. The Pacific Decadal Oscillation and Quasi Biennial Oscillation also show considerable influence, but intermittently. The El Niño Southern Oscillation, the Arctic Oscillation, and the North Atlantic Oscillation all have a weaker but interrelated influence. While the Atlantic Multidecadal Oscillation showed the weakest overall influence on regional extreme weather, it demonstrated a clear spatial gradient across the region, unlike the aforementioned oscillations. Long‐term changes in regional extreme weather are not generally important, in that a sustained increase or decrease in extreme weather events is not usually characteristic of the weather records. The primary exception to this result is for extreme minimum temperature events, whose frequency has slightly decreased since the 1880s.

To identify the extreme weather-related health needs of homeless people and the response by homeless service providers in Adelaide, South Australia, a five-phased qualitative interpretive study was undertaken. (1) Literature review, followed by semi-structured interviews with 25 homeless people to ascertain health needs during extreme weather events. (2) Identification of homeless services. (3) Semi-structured interviews with 16 homeless service providers regarding their response to the health needs of homeless people at times of extreme weather. (4) Gap analysis. (5) Suggestions for policy and planning. People experiencing homelessness describe adverse health impacts more from extreme cold, than extreme hot weather. They considered their health suffered more, because of wet bedding, clothes and shoes. They felt more depressed and less able to keep themselves well during cold, wet winters. However, homeless service providers were more focussed on planning for extra service responses during times of extreme heat rather than extreme cold. Even though a city may be considered to have a temperate climate with a history of very hot summers, primary homeless populations have health needs during winter months. The experiences and needs of homeless people should be considered in extreme weather policy and when planning responses.

Early identification of cerebral palsy (CP) is important for early intervention, yet expert-based assessments do not permit widespread use, and conventional machine learning alternatives lack validity. To develop and assess the external validity of a novel deep learning-based method to predict CP based on videos of infants' spontaneous movements at 9 to 18 weeks' corrected age. This prognostic study of a deep learning-based method to predict CP at a corrected age of 12 to 89 months involved 557 infants with a high risk of perinatal brain injury who were enrolled in previous studies conducted at 13 hospitals in Belgium, India, Norway, and the US between September 10, 2001, and October 25, 2018. Analysis was performed between February 11, 2020, and September 23, 2021. Included infants had available video recorded during the fidgety movement period from 9 to 18 weeks' corrected age, available classifications of fidgety movements ascertained by the general movement assessment (GMA) tool, and available data on CP status at 12 months' corrected age or older. A total of 418 infants (75.0%) were randomly assigned to the model development (training and internal validation) sample, and 139 (25.0%) were randomly assigned to the external validation sample (1 test set). Video recording of spontaneous movements. The primary outcome was prediction of CP. Deep learning-based prediction of CP was performed automatically from a single video. Secondary outcomes included prediction of associated functional level and CP subtype. Sensitivity, specificity, positive and negative predictive values, and accuracy were assessed. Among 557 infants (310 [55.7%] male), the median (IQR) corrected age was 12 (11-13) weeks at assessment, and 84 infants (15.1%) were diagnosed with CP at a mean (SD) age of 3.4 (1.7) years. Data on race and ethnicity were not reported because previous studies (from which the infant samples were derived) used different study protocols with inconsistent collection of these data. On external validation, the deep learning-based CP prediction method had sensitivity of 71.4% (95% CI, 47.8%-88.7%), specificity of 94.1% (95% CI, 88.2%-97.6%), positive predictive value of 68.2% (95% CI, 45.1%-86.1%), and negative predictive value of 94.9% (95% CI, 89.2%-98.1%). In comparison, the GMA tool had sensitivity of 70.0% (95% CI, 45.7%-88.1%), specificity of 88.7% (95% CI, 81.5%-93.8%), positive predictive value of 51.9% (95% CI, 32.0%-71.3%), and negative predictive value of 94.4% (95% CI, 88.3%-97.9%). The deep learning method achieved higher accuracy than the conventional machine learning method (90.6% [95% CI, 84.5%-94.9%] vs 72.7% [95% CI, 64.5%-79.9%]; P < .001), but no significant improvement in accuracy was observed compared with the GMA tool (85.9%; 95% CI, 78.9%-91.3%; P = .11). The deep learning prediction model had higher sensitivity among infants with nonambulatory CP (100%; 95% CI, 63.1%-100%) vs ambulatory CP (58.3%; 95% CI, 27.7%-84.8%; P = .02) and spastic bilateral CP (92.3%; 95% CI, 64.0%-99.8%) vs spastic unilateral CP (42.9%; 95% CI, 9.9%-81.6%; P < .001). In this prognostic study, a deep learning-based method for predicting CP at 9 to 18 weeks' corrected age had predictive accuracy on external validation, which suggests possible avenues for using deep learning-based software to provide objective early detection of CP in clinical settings.

Enhancing climate resilience in businesses: The role of artificial intelligence

Evaluation and prediction of transportation resilience under extreme weather events: A diffusion graph convolutional approach

Background:Climate change's influence on extreme weather events poses a significant threat to the morbidity and mortality of asthma patients. The aim of this study was to examine associations between extreme weather events and asthma-related outcomes.Methods:A systematic literature search for relevant studies was performed using the PubMed, EMBASE, Web of Science and ProQuest databases. Fixed-effects and random-effects models were applied to estimate the effects of extreme weather events on asthma-related outcomes.Results:We observed that extreme weather events were associated with increasing risks of general asthma outcomes with relative risks of 1.18-fold for asthma events (95% CI 1.13–1.24), 1.10-fold for asthma symptoms (95% CI 1.03–1.18) and 1.09-fold for asthma diagnoses (95% CI 1.00–1.19). Extreme weather events were associated with increased risks of acute asthma exacerbation with risk ratios of asthma emergency department visits of 1.25-fold (95% CI 1.14–1.37), of asthma hospital admissions of 1.10-fold (95% CI 1.04–1.17), of asthma outpatient visits of 1.19-fold (95% CI 1.06–1.34) and of asthma mortality of 2.10-fold (95% CI 1.35–3.27). Additionally, an increase in extreme weather events increased risk ratios of asthma events by 1.19-fold in children and 1.29-fold in females (95% CI 1.08–1.32 and 95% CI 0.98–1.69, respectively). Thunderstorms increased the risk ratio of asthma events by 1.24-fold (95% CI 1.13–1.36).Conclusions:Our study showed that extreme weather events more prominently increased the risk of asthma morbidity and mortality in children and females. Climate change is a critical concern for asthma control.

deepMDDI: A deep graph convolutional network framework for multi-label prediction of drug-drug interactions

In the research of hyperspectral image classification, deep learning-based methods have achieved promising results. However, due to the high cost and difficulty in obtaining labeled hyperspectral image samples, traditional deep learning methods experience a significant performance decline under few-shot conditions. This paper proposes a deep few-shot learning method, referred to as Wavelet-Based Multi-stream Spatial-Frequency Fusion Network, to address the issue of small sample size in hyperspectral image classification. Through a multi-stream structure, features at various scales are obtained, effectively alleviating the issue of insufficient feature representation caused by limited training samples. These features are then fed into a multi-level wavelet transformation module integrated with a frequency-domain attention module. By decomposing the input into high-frequency and low-frequency components via wavelet transformation, and subsequently inputting these components into the frequency-domain attention module, a richer representation of features in the frequency domain is achieved. Finally, a Spatial-Frequency Fusion module is employed to not only extract abundant feature information but also fuse features from different scales. This process results in feature maps that encompass more comprehensive global feature information, thereby significantly enhancing classification accuracy under small sample conditions. The proposed method has been compared with traditional machine learning approaches and recent deep learning methods on the Houston, Dioni, and University of Pavia datasets. Under training conditions using only 1 to 5, 10, 15, and 20 samples per class, the experimental results demonstrate the superiority of the method proposed in this paper.