Driver Monitoring of Automated Vehicles by Classification of Driver Drowsiness Using a Deep Convolutional Neural Network Trained by Scalograms of ECG Signals

Driver drowsiness is one of the leading causes of traffic accidents. This paper proposes a new method for classifying driver drowsiness using deep convolution neural networks trained by wavelet scalogram images of electrocardiogram (ECG) signals. Three different classes were defined for drowsiness based on video observation of driving tests performed in a simulator for manual and automated modes. The Bayesian optimization method is employed to optimize the hyperparameters of the designed neural networks, such as the learning rate and the number of neurons in every layer. To assess the results of the deep network method, heart rate variability (HRV) data is derived from the ECG signals, some features are extracted from this data, and finally, random forest and k-nearest neighbors (KNN) classifiers are used as two traditional methods to classify the drowsiness levels. Results show that the trained deep network achieves balanced accuracies of about 77% and 79% in the manual and automated modes, respectively. However, the best obtained balanced accuracies using traditional methods are about 62% and 64%. We conclude that designed deep networks working with wavelet scalogram images of ECG signals significantly outperform KNN and random forest classifiers which are trained on HRV-based features.

How does a primary school organize outdoor swimming and water safety lessons?

The main purpose of this study was to explore how a primary school organized a week with outdoor education, and especially what characterized the outdoor swimming and water safety (SWS)-lessons. The SWS-lessons were part of the physical education (PE) program. Two teachers and one headmaster from a primary school participated in the study. Four days of video observation of SWS-lessons were used in photo-elicitation interviews with teachers and the headmaster. The findings revealed that the headmaster highly prioritized the outdoor education practice in this school and that the teachers’ colleagues showed great eagerness to cooperate and prioritize these lessons. The week’s organization was characterized by collaborating management, and the outdoor SWS-lessons were characterized by experiential learning, challenge by choice, and risk awareness. The outdoor education practice corresponded well with the Norwegian curriculum goals about learning to be safe in, on, and around water.

SEISMISITY of THE VOLCANIC AREAS of KAMCHATKA in 2015

The results of near real-time monitoring of the active Kamchatka volcanoes are described. Continuous monitoring was carried out using three remote methods: 1) seismic monitoring according to automatic telemetric seismic stations; 2) visual and video observation; 3) satellite observation of the thermal anomalies and the ash clouds. Annual results of seismic activity of the Northern (Shiveluch, Kluchevskoy, Bezymianny, Krestovsky, and Ushkovsky), the Avacha (Avachinsky, and Koryaksky), the Mutnovsky-Gorely volcano groups and the Kizimen volcano are presented. 5464 earthquakes with КS=1.8–8.1 were located for the Northern volcano group, 302 earthquakes with КS=1.7–5.7 – for the Avacha volcano group, 295 earthquakes with КS=2.1–6.8 for the Mutnovsky-Gorely volcano group, 462 earthquakes with КS=2.2–8.3 for Kizimen volcano, and 165 earthquakes with КS=2.5–8.4 for Zhupanovsky volcano in 2015. Maps of epicenters, quantities of seismic energy and earthquake distribution by energy classes are given. All periods of activity were fixed and investigated by remote methods in 2015: intensive volcanic activity of the Sheveluch volcano associated with a new cone; the summit explosive-effusive eruption of the Kluchevskoy volcano in January–April; and a continuation of seismic and volcanic activity of the Zhupanovsky volcano after 56-year quite period.

Development and Psychometric Testing of Coding Scheme for Dementia Family Communication Using Video Observation

Understanding communication behaviors between persons living with dementia and family caregivers is essential for meaningful social interaction and decrease problematic behaviors and caregiving burden. The purpose of this study was to develop and test the psychometric properties of a coding scheme for dementia care interactions. The coding scheme items were developed from literature and expert review, and the pilot testing on 16 video-recorded interactions. A secondary analysis was conducted using 77 videos from 21 dyads of dementia family interactions naturally occurred in the participant’s home. The final coding scheme consists of 11 codes for persons living with dementia (6 nonverbal and 5 verbal) and 12 codes for family caregivers (7 nonverbal and 5 verbal). Content validity was excellent (I-CVI = .93, S-CVI/UA = .71, S-CVI/Ave = .93 with 6 experts). Inter-item correlation was acceptable for both caregiver codes (positive nonverbal = .21, positive verbal = .15, negative nonverbal = .36, negative verbal = .29), and patient codes (positive nonverbal = .13, positive verbal = .27, negative nonverbal = .15, negative verbal = .18). Intra-rater reliability (Cohen’s Kappa = .83, percentage of agreement = 83.88%) and inter-rater reliability (Cohen’s Kappa = .81, percentage of agreement = 81.75%) were excellent. Findings suggest the preliminary psychometric properties of the newly developed coding scheme to assess dyadic interactions of persons living with dementia and their informal caregiver in-home care situations. Future testing of the coding scheme for application in communication interventions to improve quality social interaction in dementia care is discussed.

Video observation of encounters between the automated shuttles and other traffic participants along an approach to right-hand priority T-intersection

Introduction Increasing numbers of deployment projects of automated shuttles have been taking place worldwide. Safety is one of the main concerns for their successful implementation. Therefore, it is vital to gain the knowledge about interactions between these shuttles and other traffic participants. Method Given the lack of behavioural observational studies under regular traffic conditions, the presented study applies external video recordings to explore encounters between the shuttles approaching a T-intersection and other traffic participants. The encounters of interest included a vulnerable road user in the bicycle lane, a pedestrian on the zebra crossing and a road user overtaking the shuttle. The shuttles were identified from the video by RUBA software. We analysed the encounters using T-Analyst software together with the manual observation of traffic participants' behaviour. Results From 220 h of video, 318 unique manoeuvres of the shuttle were observed and 83 encounters with other traffic participants were identified and explored. Several types of risks and behavioural patterns were identified, such as road users misusing the defensive style of the shuttles or cyclists in the bicycle lane not being sure about the shuttle’s intention. Frequent hard stops of the shuttles might be dangerous for the passengers inside and can increase the risk of rear end accidents. Conclusions The findings provide a valuable insight into the interactions between automated shuttles and other traffic participants under regular traffic conditions on one location in Oslo, Norway. The study showed that introducing automated shuttles into regular traffic can lead to the emergence of new types of interactions between the shuttles and other traffic participants.

Classification of Video Observation Data for Volcanic Activity Monitoring Using Computer Vision and Modern Neural NetWorks (on Klyuchevskoy Volcano Example)

Currently, video observation systems are actively used for volcano activity monitoring. Video cameras allow us to remotely assess the state of a dangerous natural object and to detect thermal anomalies if technical capabilities are available. However, continuous use of visible band cameras instead of special tools (for example, thermal cameras), produces large number of images, that require the application of special algorithms both for preliminary filtering out the images with area of interest hidden due to weather or illumination conditions, and for volcano activity detection. Existing algorithms use preselected regions of interest in the frame for analysis. This region could be changed occasionally to observe events in a specific area of the volcano. It is a problem to set it in advance and keep it up to date, especially for an observation network with multiple cameras. The accumulated perennial archives of images with documented eruptions allow us to use modern deep learning technologies for whole frame analysis to solve the specified task. The article presents the development of algorithms to classify volcano images produced by video observation systems. The focus is on developing the algorithms to create a labelled dataset from an unstructured archive using existing and authors proposed techniques. The developed solution was tested using the archive of the video observation system for the volcanoes of Kamchatka, in particular the observation data for the Klyuchevskoy volcano. The tests show the high efficiency of the use of convolutional neural networks in volcano image classification, and the accuracy of classification achieved 91%. The resulting dataset consisting of 15,000 images and labelled in three classes of scenes is the first dataset of this kind of Kamchatka volcanoes. It can be used to develop systems for monitoring other stratovolcanoes that occupy most of the video frame.

Driver Monitoring of Automated Vehicles by Classification of Driver Drowsiness using a Deep Convolutional Neural Network Trained by Scalograms of ECG Signals

Driver drowsiness is one of the leading causes of traffic accidents. This paper proposes a new method for classifying driver drowsiness using deep convolution neural networks trained by wavelet scalogram images of electrocardiogram (ECG) signals. Three different classes were de-fined for drowsiness based on video observation of driving tests performed in a simulator for manual and automated modes. The Bayesian optimization method is employed to optimize the hyperparameters of the designed neural networks, such as the learning rate and the number of neurons in every layer. To assess the results of the deep network method, Heart Rate Variability (HRV) data is derived from the ECG signals, some features are extracted from this data, and finally, random forest and k-nearest neighbors (KNN) classifiers are used as two traditional methods to classify the drowsiness levels. Results show that the trained deep network achieves balanced accuracies of about 77% and 79% in the manual and automated modes, respectively. However, the best obtained balanced accuracies using traditional methods are about 62% and 64%. We conclude that designed deep networks working with wavelet scalogram images of ECG signals significantly outperform KNN and random forest classifiers which are trained on HRV-based features.

Online Investigations on Optimizing the Danish Health Portal Sundhed.dk

In recent years, many people have experienced different problems and challenges in using the national Danish health portal sundhed.dk, as they find it difficult to obtain desired information about their own laboratory test results and treatment plans. Therefore, the aim of this study is to find solutions, to make patients laboratory results easily accessible and understandable for the users. To achieve this aim there will be used two participatory design methods, video observation and questionnaires. The results shows that only 43.5% normally understand their test results, whereas the remaining participants need help to understand their results.

Intraoperative dynamics of workflow disruptions and surgeons' technical performance failures: insights from a simulated operating room

Introduction Flow disruptions (FD) in the operating room (OR) have been found to adversely affect the levels of stress and cognitive workload of the surgical team. It has been concluded that frequent disruptions also lead to impaired technical performance and subsequently pose a risk to patient safety. However, respective studies are scarce. We therefore aimed to determine if surgical performance failures increase after disruptive events during a complete surgical intervention. Methods We set up a mixed-reality-based OR simulation study within a full-team scenario. Eleven orthopaedic surgeons performed a vertebroplasty procedure from incision to closure. Simulations were audio- and videotaped and key surgical instrument movements were automatically tracked to determine performance failures, i.e. injury of critical tissue. Flow disruptions were identified through retrospective video observation and evaluated according to duration, severity, source, and initiation. We applied a multilevel binary logistic regression model to determine the relationship between FDs and technical performance failures. For this purpose, we compared FDs in one-minute intervals before performance failures with intervals without subsequent performance failures. Results Average simulation duration was 30:02 min (SD = 10:48 min). In 11 simulated cases, 114 flow disruption events were observed with a mean hourly rate of 20.4 (SD = 5.6) and substantial variation across FD sources. Overall, 53 performance failures were recorded. We observed no relationship between FDs and likelihood of immediate performance failures: Adjusted odds ratio = 1.03 (95% CI 0.46–2.30). Likewise, no evidence could be found for different source types of FDs. Conclusion Our study advances previous methodological approaches through the utilisation of a mixed-reality simulation environment, automated surgical performance assessments, and expert-rated observations of FD events. Our data do not support the common assumption that FDs adversely affect technical performance. Yet, future studies should focus on the determining factors, mechanisms, and dynamics underlying our findings.