Search

Aim This study aimed to develop an AI model for detecting a caller’s emotional state during out-of-hospital cardiac arrest calls by processing audio recordings of dispatch communications. Methods Audio recordings of 337 out-of-hospital cardiac arrest calls from March-April 2011 were retrieved. The callers’ emotional state was classified based on the emotional content and cooperative scores. Mel-frequency cepstral coefficients extracted essential information from the voice signals. A support vector machine was utilised for the automatic judgement, and repeated random sub-sampling cross validation (RRS-CV) was applied to evaluate robustness. The results from the artificial intelligence classifier were compared with the consensus of expert reviewers. Results The audio recordings were classified into five emotional content and cooperative score levels. The proposed model had an average positive predictive value of 72.97%, a negative predictive value of 93.47%, sensitivity of 38.76%, and specificity of 98.29%. If only the first 10 seconds of the recordings were considered, it had an average positive predictive value of 84.62%, a negative predictive value of 93.57%, sensitivity of 52.38%, and specificity of 98.64%. The artificial intelligence model’s performance maintained preferable results for emotionally stable cases. Conclusion Artificial intelligence models can possibly facilitate the judgement of callers’ emotional states during dispatch conversations. This model has the potential to be utilised in practice, by pre-screening emotionally stable callers, thus allowing dispatchers to focus on cases that are judged to be emotionally unstable. Further research and validation are required to improve the model's performance and make it suitable for the general population.

Travel demands in remote areas are scattered and unstable. Thus, the integration of different types of transportation can be beneficial for improving the service level of transportation systems in ...

Understanding the associations between crash characteristics facilitates the development of traffic safety policies for improving traffic safety. This study investigates the temporal stability of associations between crash characteristics at different temporal levels using multiple correspondence analysis (MCA). For each date in 2020, crash data from the previous week, month, season, half year, one year, two years, three years, and four years are collected respectively as eight temporal levels. MCA plots and chi-square distance analysis are used to assess the temporal stability of associations between crash characteristics across dates in 2020 with data from various temporal levels. The key findings of this study demonstrate that associations between crash characteristics at lower temporal levels show notable and potential cyclical variations across dates, while more stable and long-term trend of associations between crash characteristics may be identified as the temporal level increases, especially at the two-year level and higher temporal levels at which temporal stability may be expected. The study contributes to the literature by presenting a challenge for traffic analysts in that both temporally stable and unstable associations between crash characteristics may be observed at any point in time when different temporal levels are considered as study periods. Therefore, it may serve as a foundation for future research and practical works to identify traffic safety issues and optimal policies as well as facilitate the interpretation of statistical modeling in the presence of temporally unstable data.

This research estimated the spatial-temporal distribution of humans in buildings through image sensing. Inputs were the in-building network, image sequences recording the movement of human,...

BACKGROUND Early recognition of severely injured patients in prehospital settings is of paramount importance for timely treatment and transportation of patients to further treatment facilities. The dispatching accuracy has seldom been addressed in previous studies. OBJECTIVE In this study, we aimed to build a machine learning–based model through text mining of emergency calls for the automated identification of severely injured patients after a road accident. METHODS Audio recordings of road accidents in Taipei City, Taiwan, in 2018 were obtained and randomly sampled. Data on call transfers or non-Mandarin speeches were excluded. To predict cases of severe trauma identified on-site by emergency medical technicians, all included cases were evaluated by both humans (6 dispatchers) and a machine learning model, that is, a prehospital-activated major trauma (PAMT) model. The PAMT model was developed using term frequency–inverse document frequency, rule-based classification, and a Bernoulli naïve Bayes classifier. Repeated random subsampling cross-validation was applied to evaluate the robustness of the model. The prediction performance of dispatchers and the PAMT model, in severe cases, was compared. Performance was indicated by sensitivity, specificity, positive predictive value, negative predictive value, and accuracy. RESULTS Although the mean sensitivity and negative predictive value obtained by the PAMT model were higher than those of dispatchers, they obtained higher mean specificity, positive predictive value, and accuracy. The mean accuracy of the PAMT model, from certainty level 0 (lowest certainty) to level 6 (highest certainty), was higher except for levels 5 and 6. The overall performances of the dispatchers and the PAMT model were similar; however, the PAMT model had higher accuracy in cases where the dispatchers were less certain of their judgments. CONCLUSIONS A machine learning–based model, called the PAMT model, was developed to predict severe road accident trauma. The results of our study suggest that the accuracy of the PAMT model is not superior to that of the participating dispatchers; however, it may assist dispatchers when they lack confidence while making a judgment.

Background Early recognition of severely injured patients in prehospital settings is of paramount importance for timely treatment and transportation of patients to further treatment facilities. The dispatching accuracy has seldom been addressed in previous studies. Objective In this study, we aimed to build a machine learning–based model through text mining of emergency calls for the automated identification of severely injured patients after a road accident. Methods Audio recordings of road accidents in Taipei City, Taiwan, in 2018 were obtained and randomly sampled. Data on call transfers or non-Mandarin speeches were excluded. To predict cases of severe trauma identified on-site by emergency medical technicians, all included cases were evaluated by both humans (6 dispatchers) and a machine learning model, that is, a prehospital-activated major trauma (PAMT) model. The PAMT model was developed using term frequency–inverse document frequency, rule-based classification, and a Bernoulli naïve Bayes classifier. Repeated random subsampling cross-validation was applied to evaluate the robustness of the model. The prediction performance of dispatchers and the PAMT model, in severe cases, was compared. Performance was indicated by sensitivity, specificity, positive predictive value, negative predictive value, and accuracy. Results Although the mean sensitivity and negative predictive value obtained by the PAMT model were higher than those of dispatchers, they obtained higher mean specificity, positive predictive value, and accuracy. The mean accuracy of the PAMT model, from certainty level 0 (lowest certainty) to level 6 (highest certainty), was higher except for levels 5 and 6. The overall performances of the dispatchers and the PAMT model were similar; however, the PAMT model had higher accuracy in cases where the dispatchers were less certain of their judgments. Conclusions A machine learning–based model, called the PAMT model, was developed to predict severe road accident trauma. The results of our study suggest that the accuracy of the PAMT model is not superior to that of the participating dispatchers; however, it may assist dispatchers when they lack confidence while making a judgment.

Vehicle emissions are an important source of particulate matter (PM) in urban areas and have well-known adverse health effects on human health. Oxidative potential (OP) is used as a quantification metric for indexing PM toxicity. In this study, by using a liquid spot sampler (LSS) and the dithiothreitol (DTT) assay, the diurnal OP variation was assessed at a ground-level urban monitoring station. Besides, since the monitoring station was adjacent to the main road, the correlation between OP and traffic volume was also evaluated. PM components, including metals, water-soluble inorganic aerosols (WSIAs), black carbon (BC), and polycyclic aromatic hydrocarbons (PAHs), were also simultaneously monitored. The daytime and evening mean ± std volume-normalized OP (OPv) were 0.46 ± 0.27 and 0.48 ± 0.26 nmol/min/m3, and exhibited good correlations with PM1.0 and BC; however, these concentrations were only weakly correlated with mass-normalized OP (OPm). The mean ± std OPm was higher in the daytime (41.3 ± 13.8 pmol/min/μg) than in the evening (36.1 ± 11.5 pmol/min/μg). According to the PMF analysis, traffic emissions dominated the diurnal OP contribution. Organic matter and individual metals associated with non-exhaust traffic emissions, such as Mn, Fe, and Cu, contributed substantially to OP. Diurnal variations of PAH concentrations suggest that photochemical reactions could enhance OP, highlighting the importance of atmospheric aging on PM toxicity.

The installation of automated external defibrillators (AEDs) in public places is encouraged, and the determination of the installation locations could be formed into mathematical programming models. In previous studies, data and parameters for those models were generated experimentally or manually. To estimate the parameters in practice for better reliability of the model outcomes, this research aims to design a procedure for the estimation of the real human demand distribution through automatic approaches. Inputs of the proposed work include the image sequences taken inside the building and the network describing the in-building geometry. Images are analyzed via computer vision methods for detection and tracking of people, and also augmented with the extra dimension of image depth estimation through a deep convolutional neural network. The key indicator to project the image back to the real space is identified by the correspondence between the location of humans appear in the image and the location of intersections of camera field of view with the network. Finally, the spatial-temporal human count is recorded to be the output and could serve as input parameter for related AED installation problems. The process is validated in image sequences taken at a corridor in a school building; the performance of each step and the influence of each step is discussed. In conclusion, our research provides a more automatic and robust approach to obtain the real spatial-temporal human distribution in public buildings.

Transcription initiation requires formation of the open promoter complex, RPo. To generate RPo, RNA polymerase (RNAP) unwinds the DNA duplex to form the transcription bubble and loads the DNA into the RNAP active site. RPo formation is a multi-step process with transient intermediates of unknown structure. We used single particle cryo-electron microscopy to visualize seven intermediates containing Escherichia coli RNAP with the transcription factor TraR en route to forming RPo. The structures span the RPo formation pathway from initial recognition of the duplex promoter in a closed complex to the final RPo. The structures and supporting biochemical data define RNAP and promoter DNA conformational changes that delineate steps on the pathway, including previously undetected transient promoter-RNAP interactions that contribute to populating the intermediates but do not occur in RPo. Our work provides a structural basis for understanding RPo formation and its regulation, a major checkpoint in gene expression throughout evolution.

The stringent response to nutrient deprivation is a stress response found throughout the bacterial domain of life. Although first described in proteobacteria for matching ribosome synthesis to the cell's translation status and for preventing formation of defective ribosomal particles, the response is actually much broader, regulating many hundreds of genes—some positively, some negatively. Utilization of the signaling molecules ppGpp and pppGpp for this purpose is ubiquitous in bacterial evolution, although the mechanisms employed vary. In proteobacteria, the signaling molecules typically bind to two sites on RNA polymerase, one at the interface of the β′ and ω subunits and one at the interface of the β′ secondary channel and the transcription factor DksA. The β′ secondary channel is targeted by other transcription regulators as well. Although studies on the transcriptional outputs of the stringent response date back at least 50 years, the mechanisms responsible are only now coming into focus.

A methodology for optimizing variable pedestrian evacuation guidance in buildings with convex polygonal interior spaces is proposed. The optimization of variable guidance is a bi-level problem. The calculation of variable guidance based on the prediction of congestion and hazards is the upper-level problem. The prediction of congestion provided the variable guidance is the lower-level problem. A local search procedure is developed to solve the problem. The proposed methodology has three major contributions. First, a logistic regression model for guidance compliance behavior is calibrated using a virtual reality experiment and the critical factors for the behavior are identified. Second, the guidance compliance and following behaviors are considered in the lower-level problem. Third, benchmarks are calculated to evaluate the performance of optimized variable guidance, including the lower bound of the maximum evacuation time and the maximum evacuation time under a fixed guidance. Finally, the proposed methodology is validated with numerical examples. Results show that the method has the potential to reduce evacuation time in emergencies.

TraR and its homolog DksA are bacterial proteins that regulate transcription initiation by binding directly to RNA polymerase (RNAP) rather than to promoter DNA. Effects of TraR mimic the combined effects of DksA and its cofactor ppGpp, but the structural basis for regulation by these factors remains unclear. Here, we use cryo-electron microscopy to determine structures of Escherichia coli RNAP, with or without TraR, and of an RNAP-promoter complex. TraR binding induced RNAP conformational changes not seen in previous crystallographic analyses, and a quantitative analysis revealed TraR-induced changes in RNAP conformational heterogeneity. These changes involve mobile regions of RNAP affecting promoter DNA interactions, including the βlobe, the clamp, the bridge helix, and several lineage-specific insertions. Using mutational approaches, we show that these structural changes, as well as effects on σ70 region 1.1, are critical for transcription activation or inhibition, depending on the kinetic features of regulated promoters., eLife digest Cells need to make proteins in order to survive. To make a protein, a cell first needs to take the information from a gene coded in its DNA and copy it into a template made of a different molecule called RNA via a process called transcription. Not all genes are transcribed or expressed at the same time or the same rate, because a cell needs different proteins depending on its environment. The molecules that regulate when each gene is transcribed are called transcription factors. When the bacterium Escherichia coli is starved for the building blocks it needs to make proteins, it changes the expression of almost one quarter of its genes within 5 minutes. This broad response requires two transcription factors, ppGpp and DksA. Unlike most transcription factors, these two molecules bind directly to the enzyme responsible for transcribing DNA into RNA, the RNA polymerase, but they do not bind to DNA. Another transcription factor called TraR can mimic the combined effects of ppGpp and DksA on transcription, changing the conformation of RNA polymerase in the same way. Now, Chen et al. have used a high-resolution imaging technique called cryo-electron microscopy to reveal the details of how the structure of RNA polymerase changes in response to TraR binding, and by analogy, in response to ppGpp and DksA. The experiments showed that TraR interacts with regions of RNA polymerase that move when the protein binds DNA. These interactions can either help or hinder the start of transcription depending on which DNA sequence the RNA polymerase binds. Using the structures obtained via cryo-electron microscopy as guides, Chen et al. mutated each of the mobile parts of RNA polymerase individually to determine which interactions with TraR were necessary to change gene expression. These results shed light on a fundamental process in all living cells, the initiation of transcription, and how it changes in response to the cell's nutritional environment. This may help explain how different cells regulate gene expression and may also lead to the development of new antibiotics.

TraR and its homolog DksA are bacterial proteins that regulate transcription initiation by binding directly to RNA polymerase (RNAP) rather than to promoter DNA. Effects of TraR mimic the combined effects of DksA and its cofactor ppGpp. How TraR and its homologs regulate transcription is unclear. Here, we use cryo-electron microscopy to determine structures ofEscherichia coliRNAP, with or without TraR, and of an RNAP-promoter complex. TraR binding induced RNAP conformational changes not seen in previous crystallographic analyses, and a quantitative analysis of RNAP conformational heterogeneity revealed TraR-induced changes in RNAP dynamics. These changes involve mobile regions of RNAP affecting promoter DNA interactions, including the βlobe, the clamp, the bridge helix, and several lineage-specific insertions. Using mutational approaches, we show that these structural changes, as well as effects on σ70region 1.1, are critical for transcription activation or inhibition, depending on the kinetic features of regulated promoters.

This research provides two main changes based on Detect-And-Track. To improve the Multi-Object Tracking Accuracy (MOTA) while keeping the lightweight of the original approach, this paper proposes a gradient approach to obtain higher MOTA. We use the location of two previous frames of the same identified person to calculate the gradient for the location prediction of the current frame. Then, the predicted and the detected locations are compared. We also compare the current and previous detections. With a weighted combination for matching, we increase the MOTA score and improve the results of Detect-And-Track. Moreover, this research replaces cosine distance, the original feature extractor, with Euclidean distance. By doing so, feature extraction can match Intersection over Union (IoU) better. The weighted combination, which consists of IoU and Euclidean distance, provides a better MOTA than Detect-And-Track. In addition, a greedy approach facilitates a higher MOTA when implement with IoU and Euclidean distance. This weighted combination utility is superior than the combination of IoU and cosine distance, achieving 56.1% MOTA in total on the validation data of PoseTrack ICCV’17 dataset.

Unpredictable and no-notice disasters such as flash floods and wildfires, have been increasing. Under such circumstances, large-scale evacuation is required and the population may not be able to be effectively relocated in time from the disaster-threatened zone. The evacuees may be risking their lives during evacuation traveling toward the designated shelters/safety zones. A Shelter-In-Place (SIP) strategy is proposed to provide immediate avoidance of hazards for communities. This study develops the optimization models to determine SIP deployment and evacuee assignment for the case of vertical evacuation (VE) in response to dam-break floods. Compared with general shelters located outside evacuation areas, SIPs are designed to be reachable on foot upon the occurrence of sudden disasters. Accessibility of SIPs is one of the primary concerns in the problem context of determining their locations. Multi-class evacuees are addressed in the SIP strategy, particularly for the elderly in consideration of their vulnerability in disasters and heterogeneous walking speed across different classes. Additionally, the spatially-varying risk over the disaster-threatened areas is also considered in the model. A dam failure scenario causing flooding is studied for Tucheng District, New Taipei City, Taiwan. The case study illustrates the effects of spatial risk consideration, elderly priority, and coverage distance on the optimal solution of shelter locations. The extent of improvement in emergency planning is also fundamentally related to the population distribution in the study area.

Abnormal activities on construction jobsites may compromise productivity and pose threat to workers' safety. This paper proposes the analysis of consecutive image sequences for automatic identification of irregular operations and their visualization. The data analytics is composed of four steps: object detection, object tracking, action recognition, and operational analysis. The Faster Region-proposal Convolutional Neural Network (Faster R-CNN) is adapted with transfer learning for detection of workers and pieces of construction equipment on the jobsite, while the Simple Online and Realtime Tracking (SORT) approach is applied for object tracking. A hybrid model integrating CNN and Long Short Term Memory (LSTM) is employed for action recognition. An alternative form of the Crew-balance Chart (CBC), called line chart in which anomalies are pre-screened, is utilized for recognized actions. Validation was carried out with earthmoving operations. The trained Faster R-CNN reached a 73% Average Precision (AP), and the SORT algorithm modified by this work successfully reduced identity switches. Irregular operations in the testing videos were identified, and truck exchanges were filtered. In addition, an activity log was produced with basic information along with starting and ending times of the identified irregular operations. With the line chart and the log provided by the proposed framework, field managers can efficiently identify potential abnormal activities, providing the opportunity for further investigations and adjustments accordingly.

Pre-hospital Emergency Medical Service (EMS) provides the immediate and appropriate aid to patients in emergencies. As part of the traditional triad of first responders, EMS plays an important role in disaster response. In this work, the transportation infrastructure, which the EMS is dependent on, is considered. The objective of this research is to improve the effectiveness of EMS after the disaster by applying integer programming and the network-based partitioning to determine temporary locations for on-post EMS facilities. Integer Programming problems are formed for the optimization problem in different scales, and the Lagrangian Relaxation is adapted to extend the problem further into larger scale. Network based partitioning of demands are also proposed and tested. Numerical results are provided, and a case study is presented. In the case study, the facility location problem takes into consideration of both disaster triggered and usual EMS demand that forms a worst-case scenario. The analytical results are expected to facilitate decision making, and to serve as benchmarks for the planning of post-disaster EMS.

Travel time prediction has been the fundamental brick for developing both Advanced Traffic Management Systems (ATMS) and Advanced Traveler Information Systems (ATIS). The problem of travel time prediction is primarily shaped by the data available for model development and the needs from the perspectives of both travelers and operators. A distance-based Electronic Toll Collection (ETC) system over the freeway system in Taiwan has been fully implemented since December 2013, which provides more comprehensive online data acquisition of freeway traffic conditions, thereby enabling the capability to more reliably predict of travel time over freeway segments. Based on the data from both the ETC system and traditional Vehicle Detectors (VDs), this research proposes a travel time prediction approach whose core technique of data fusion seeks to seamlessly capture the spatio-temporal pattern of freeway traffic flows by matching traffic dynamics revealed from the ETC and VD data. Further, prediction models are constructed thereupon, where the Kalman filter is employed for short-term prediction and the Fourier transform for long-term prediction based on the continuous parameterized modeling of spot travel speed. The proposed approach is implemented as an online deployable system using Java, together with real-world data collected from the freeway in Taiwan, for the numerical experiments. The prediction errors are no greater than 10% in most cases, which illustrates the high accuracy of prediction capability of the model. The encouraging results also highlight the benefits of the pre-processing of data and data fusion in improving data quality and applicability.

Multiple construction worker tracking is an active research area critical to the planning of the job site. Challenges in multiple construction worker tracking include miss detection and mismatch due to occlusion and identity switches. To the best knowledge of the authors, the mismatch is not reported in the literature of construction for image based single camera multiple worker tracking. As a result, the mismatch should be taken into account through a representative performance index such as the Multi-Object Tracking Accuracy (MOTA). This work aims to improve the performance of the current multiple worker tracking through an approach composed of three stages: detection, matching and re-matching. In the detection stage, the deep learning detector, Mask R-CNN, is utilized. In the matching stage, we attempt to track workers between consecutive image frames through a gradient based method with feature based comparison. Several cost means and matching methods have been experimented for model selection. Trajectories of tracking objects are derived in this stage. The best cost measurements and matching methods are recommended. Trajectories of tracking objects could be interrupted because of miss detection or mismatch. We call those broken trajectories, without matched detections, orphans. In the re-matching stage, we attempt to recover unmatched detections in the current frame with previous orphans based on extracted features. A competitive MOTA of 56.7% was obtained from the proposed approach over MOTA of 55.9% from the state-of-the-art Detect-And-Track model on a human tracking benchmark dataset. On construction job sites, we have tested the approach with 4 testing videos, resulting in a total MOTA of 81.8%, average MOTA per video of 79.0% and standard deviation of 13.0%, while the maximum and minimum MOTAs are 96.0% and 69.0%, respectively. As a result, the proposed work could potentially provide better multiple worker tracking on the construction job site. Additionally, to have a better representation of the tracking errors, this work suggests to utilize the MOTA for multiple construction worker tracking.

The mixed traffic flow has complex dynamics by nature. The kinematic differences between automobiles and motorcycles result to distinct driving behaviors. Traditional automobile-based traffic flow theory is not always suitable for mixed traffic streams. The purpose of this study is to observe from actual data a clearance boundary, called Safety Space, drivers maintain from other vehicles, and use it as a spatial filter to determine conflicts in mixed traffic flows. Image data are collected from an Unmanned Aerial Vehicle (UAV), and microscopic characteristics such as vehicle type, position, velocity, and trajectory are extracted through computer vision techniques. The Histogram of Oriented Gradients (HOG) feature and the Support Vector Machine (SVM) classifier are utilized for the vehicle detection, while the Kalman Filter is employed for the derivation of vehicle trajectories. The Safety Space is then determined based on those trajectories. Validation data are collected at intersections in Taipei, Taiwan; Bangkok, Thailand; and Mumbai, India. The vehicle detection and tracking are satisfactory, and the Safety Space surrogate reveals risk zones caused by spatial proximity between vehicles.

The Escherichia coli F element-encoded protein TraR is a distant homolog of the chromosome-encoded transcription factor DksA. Here we address the mechanism by which TraR acts as a global regulator, inhibiting some promoters and activating others. We show that TraR regulates transcription directly in vitro by binding to the secondary channel of RNA polymerase (RNAP) using interactions similar, but not identical, to those of DksA. Even though it binds to RNAP with only slightly higher affinity than DksA and is only half the size of DksA, TraR by itself inhibits transcription as strongly as DksA and ppGpp combined and much more than DksA alone. Furthermore, unlike DksA, TraR activates transcription even in the absence of ppGpp. TraR lacks the residues that interact with ppGpp in DksA, and TraR binding to RNAP uses the residues in the β′ rim helices that contribute to the ppGpp binding site in the DksA–ppGpp–RNAP complex. Thus, unlike DksA, TraR does not bind ppGpp. We propose a model in which TraR mimics the effects of DksA and ppGpp together by binding directly to the region of the RNAP secondary channel that otherwise binds ppGpp, and its N-terminal region, like the coiled-coil tip of DksA, engages the active-site region of the enzyme and affects transcription allosterically. These data provide insights into the function not only of TraR but also of an evolutionarily widespread and diverse family of TraR-like proteins encoded by bacteria, as well as bacteriophages and other extrachromosomal elements.