In conditionally automated driving (SAE level 3), drivers may take their eyes off the road but will still need to be ready to take control and will, therefore, benefit from information on automation. This study aims to investigate the effectiveness of automation manoeuvre information provided through spatial sound, traditional notification sound (beep), and a visual interface. Spatial sounds were designed differentiating four distinct driving manoeuvres: overtaking a leading car, slowing down, turning right, and passing a roundabout. The notification sound consisted of one beep being identical for all manoeuvres. The visual interface showed the automation mode with an image and manoeuvre information with text and images. The impact of these interfaces on trust, workload, acceptance, situation awareness, and sense of control was evaluated with questionnaires and visual attention was evaluated with eye tracking while participants engaged in a visual-motor secondary task in a driving simulator. The results indicate that, with all interfaces tested, manoeuvre information enhances trust, acceptance, situation awareness, and sense of control, without significantly affecting the overall workload. These benefits were more profound, adding auditory information and differed marginally between the traditional notification and the spatial sound, as the effectiveness of the different auditory interface types varied depending on the specific manoeuvre. Findings highlight the importance of designing user interfaces for automation manoeuvre information using auditory cues to improve the user experience in automated driving.@en

Trust and perceived safety are pivotal in the acceptance of automated vehicles and can be enhanced by providing users with automation information on the (safe) operation of the vehicle. This study aims to identify how user interfaces (UI) can enhance drivers' trust and acceptance and reduce perceived risk in partially automated vehicles. Four interfaces were designed with different levels of complexity. These levels were achieved by combining automation information (surrounding information vs surrounding and manoeuvre information) and modality (visual vs visual and auditory). These interfaces were evaluated in a driving simulator in which a partially automated vehicle reacted to an event of a merging and braking vehicle in its front. The criticality of the events was manipulated by the factors merging gap (in meters) and deceleration (m/s²) of the vehicle in front. The reaction of the automation was either to brake or to change lanes. The results show that an optimal combination of automation information and modality enhances drivers' trust and acceptance. More specifically, the most advanced UI, which provided surrounding and manoeuvre information via the visual and auditory modalities, was associated with the highest trust and acceptance ranking and the lowest perceived risk. Manoeuvre information delivered through the auditory modality was particularly effective in enhancing trust and acceptance. The benefits of the UIs were consistent over events. However, in the most critical events, drivers did not feel entirely safe and did not trust the automation completely. This study suggests that the design of UIs for partially automated vehicles shall include automation information via visual and auditory modalities.

@en

Intensive Care Unit (ICU) nurses are burdened by excessive number of false and irrelevant alarms generated by patient monitoring systems. Nurses rely on these patient monitoring systems for timely and relevant medical information concerning patients. However, the systems currently in place are not sensitive to the perceptual and cognitive abilities of nurses and thus fail to communicate information efficiently. An efficient communication and an effective collaboration between patient monitoring systems and ICU nurses is only possible by designing systems sensitive to the abilities and preferences of nurses. In order to design these sensitive systems, we need to gain in-depth understanding of the user group through revealing their latent individual characteristics. To this end, we conducted a survey on individual characteristics involving nurses from two IC units. Our results shed light on the personality and other characteristics of ICU nurses. Subsequently, we performed hierarchical cluster analysis to develop data-driven nurse profiles. We suggest design recommendations tailored to four distinct user profiles to address their unique needs. By optimizing the system interactions to match the natural tendencies of nurses, we aspire to alleviate the cognitive burden induced by system use to ensure that healthcare providers receive relevant information, ultimately improving patient safety.

@en

We present a systematic approach for training and testing structural texture similarity metrics (STSIMs) so that they can be used to exploit texture redundancy for structurally lossless image compression. The training and testing is based on a set of image distortions that reflect the characteristics of the perturbations present in natural texture images. We conduct empirical studies to determine the perceived similarity scale across all pairs of original and distorted textures. We then introduce a data-driven approach for training the Mahalanobis formulation of STSIM based on the resulting annotated texture pairs. Experimental results demonstrate that training results in significant improvements in metric performance. We also show that the performance of the trained STSIM metrics is competitive with state of the art metrics based on convolutional neural networks, at substantially lower computational cost.

@en

The contour depth methodology enables non-parametric summarization of contour ensembles by extracting their representatives, confidence bands, and outliers for visualization (via contour boxplots) and robust downstream procedures. We address two shortcomings of these methods. Firstly, we significantly expedite the computation and recomputation of Inclusion Depth (ID), introducing a linear-time algorithm for epsilon ID, a variant used for handling ensembles with contours with multiple intersections. We also present the inclusion matrix, which contains the pairwise inclusion relationships between contours, and leverage it to accelerate the recomputation of ID. Secondly, extending beyond the single distribution assumption, we present the Relative Depth (ReD), a generalization of contour depth for ensembles with multiple modes. Building upon the linear-time eID, we introduce CDclust, a clustering algorithm that untangles ensemble modes of variation by optimizing ReD. Synthetic and real datasets from medical image segmentation and meteorological forecasting showcase the speed advantages, illustrate the use case of progressive depth computation and enable non-parametric multimodal analysis. To promote research and adoption, we offer the contour-depth Python package.

@en

Ensembles of contours arise in various applications like simulation, computer-Aided design, and semantic segmentation. Uncovering ensemble patterns and analyzing individual members is a challenging task that suffers from clutter. Ensemble statistical summarization can alleviate this issue by permitting analyzing ensembles' distributional components like the mean and median, confidence intervals, and outliers. Contour boxplots, powered by Contour Band Depth (CBD), are a popular non-parametric ensemble summarization method that benefits from CBD's generality, robustness, and theoretical properties. In this work, we introduce Inclusion Depth (ID), a new notion of contour depth with three defining characteristics. First, ID is a generalization of functional Half-Region Depth, which offers several theoretical guarantees. Second, ID relies on a simple principle: The inside/outside relationships between contours. This facilitates implementing ID and understanding its results. Third, the computational complexity of ID scales quadratically in the number of members of the ensemble, improving CBD's cubic complexity. This also in practice speeds up the computation enabling the use of ID for exploring large contour ensembles or in contexts requiring multiple depth evaluations like clustering. In a series of experiments on synthetic data and case studies with meteorological and segmentation data, we evaluate ID's performance and demonstrate its capabilities for the visual analysis of contour ensembles.

@en

Mode awareness is important for the safe use of automated vehicles, yet drivers' understanding of mode transitions has not been sufficiently investigated. In this study, we administered an online survey to 838 respondents to examine their understanding of control responsibilities in partial and conditional driving automation with four types of interventions (brake pedal, steering wheel, gas pedal, and take-over request). Results show that most drivers understand that they are responsible for speed and distance control after brake pedal interventions and steering control after steering wheel interventions. However, drivers have mixed responses regarding the responsibility for speed and distance control after steering wheel interventions and the responsibility for steering control after gas pedal interventions. With a higher automation level (conditional driving automation), drivers expect automation to remain responsible more often compared to a lower automation level (partial driving automation). Regarding Hands-on requirements, more than 99% of respondents answered that drivers would keep their hands on the steering wheel after all intervention types in partial automation, while 60–95% would place their hands on the wheel after various intervention types in conditional automation. A misalignment between actual logic and drivers' expectations regarding control responsibilities is observed by comparing survey responses to the mode transition logic of commercial partially automated vehicles. To resolve confusion about control responsibilities and ensure consistent expectations, we propose implementing a consistent mode design and providing enhanced information to drivers.

@en

Artificial Intelligence (AI)-based auto-delineation technologies rapidly delineate multiple structures of interest like organs-at-risk and tumors in 3D medical images, reducing personnel load and facilitating time-critical therapies. Despite its accuracy, the AI may produce flawed delineations, requiring clinician attention. Quality assessment (QA) of these delineations is laborious and demanding. Delineation error detection systems aim to aid QA, yet questions linger about clinician adoption, challenges, and time-saving potential. In this study, we address these queries in two stages. First, we investigate the error detection workflow of a radiotherapy technologist and a radiation oncologist from Holland Proton Therapy Center, a Dutch cancer treatment center. The user study revealed which information sources clinicians prefer to use for the error prioritization task and elucidated clinicians' slice-based navigation workflows with and without system assistance. Based on the findings from the user study, we developed a simulation model of the QA process, which we used to assess different error detection workflows on a retrospective cohort of 42 head and neck cancer patients. The simulation study results indicate potential time savings through error and dose information, contingent on per-slice analysis time remaining near the current baseline. Our findings encourage the development of user-centric delineation error detection systems and provide a new way to model and evaluate these systems' potential clinical value.@en

Automation of vehicles not only provides greater safety, but also many previously unimagined opportunities, such as less inequality, less stress, and more meaningful activities while driving. However, the uptake and implementation of automated driving have been falling short of its promise, due to challenges in identifying safe and acceptable ways for humans to interact with automated vehicles. The human-machine interface (HMI) in vehicles plays a more critical role today than ever before. The main research question addrssed in this study, as part of the EU-funded HADRIAN project, is: What steps need to be taken to achieve effective fluid HMI (f-HMI) design for improved driver-vehicle dialogue, especially in critical scenarios?.

@en

The design of take-over requests in automated vehicles traditionally focuses on safety and reaction time. We are interested in how take-over requests can be designed from a broader user experience perspective while ensuring safety. This paper proposes designs for ambient sound (i.e., soundscape) and driving noise to inform the driver of transition situations. Drivers must take-over control within the time budget, the time from the take-over request to the automation system limit. The time required for a safe transition depends on the complexity of the driving environment. In a scheduled take-over, which is not an emergency, there is an opportunity for an interaction that gradually introduces the driver into the transition process. Ambient sound is expected to lead the driver back to the loop with comfort, creating a novel transition experience as well as safety.

@en

Human-Machine Interfaces (HMIs) aim to support the interaction between automated vehicles and drivers to improve safety and driver experience. With the development of automated vehicles, drivers interact with vehicles in new scenarios. In addition to visual modality, sound is the other modality often used in vehicles. Previously, sounds were mainly used for alarms, but they can be used in other ways in automated vehicles. Therefore, a new approach to sound design is needed. We proposed an interactive approach for sound design to improve driver safety and user experience in automated vehicles. In this study, we suggested that the driver's interaction with automated vehicles should be analyzed based on the user and contextual understanding, and the sound should be designed to consider the appropriateness of situation matching and alert levels. This study showed that the approach supports designing sounds that enhance vehicle and driver interaction.@en

Nurses rely on patient monitoring systems for care delivery in ICUs. Monitoring systems communicate information to nurses and alert them through audiovisual alarms. However, excessive numbers of alarms often interrupt nurses in their tasks, and desensitize them to alarms. The affective consequence of this problem is that nurses are annoyed and feel frustration towards monitoring alarms. This situation leads to stress on nurses and threatens patient safety. Literature on sound annoyance distinguishes between annoyance induced by bottom-up (perceptual) and top-down (cognitive) processing. Extensive research on perceptual annoyance informs us on how to alleviate the problem by better sound design. However, addressing the cognitive aspect requires a broader understanding of annoyance as a construct. To this end, in this paper we distinguish between the annoyance induced by sensory unpleasantness of alarm sounds, and annoyance induced by frequent task interruptions. We present a conceptual framework in which we can interpret nurses’ annoyance by monitoring alarms. We further present descriptive analysis of the occurrence frequency of patient monitoring alarms in a neonatal ICU to illustrate the current state with regards to alarms. We aim to support nurses’ organizational well- being by providing an alternative hypothesis to explaining nurses’ affective states caused by auditory alarms. Future research can benefit from this paper through understanding of the context and familiarizing with the cognitive processes relevant to processing of patient monitoring alarms.@en

Delineation of tumours and organs-at-risk permits detecting and correcting changes in the patients' anatomy throughout the treatment, making it a core step of adaptive external beam radiotherapy. Although auto-contouring technologies have sped up this process, the time needed to perform the quality assessment of the generated contours remains a bottleneck, taking clinicians between several minutes and an hour to complete. The authors of this article conducted several interviews and an observational study at two treatment centres in the Netherlands to identify challenges and opportunities for speeding up the delineation process in adaptive therapies. The study revealed three contextual variables that influence contouring performance: usable additional information, applicable domain-specific knowledge, and available editing capabilities in contouring software. In practice, clinicians leverage these variables to accelerate contouring in two ways. First, they use domain-specific knowledge and relevant clinical features such as the proximity of the organs-at-risk to the tumour to enable targeted inspection of the delineation. Second, clinicians modulate editing precision depending on the effect they anticipate the edit will have on the patient outcome. By implementing these acceleration strategies in guidelines and contouring tools, developers and workflow builders could increase contouring efficiency and consistency without affecting the patient outcome.@en

Delineation of tumors and organs-at-risk permits detecting and correcting changes in the patients' anatomy throughout the treatment, making it a core step of adaptive proton therapy (APT). Although AI-based auto-contouring technologies have sped up this process, the time needed to perform the quality assessment (QA) of the generated contours remains a bottleneck, taking clinicians between several minutes up to an hour to complete. This paper introduces a fast contouring workflow suitable for time-critical APT, enabling detection of anatomical changes in shorter time frames and with a lower demand of clinical resources. The proposed AI-infused workflow follows two principles uncovered after reviewing the APT literature and conducting several interviews and an observational study in two radiotherapy centers in the Netherlands. First, enable targeted inspection of the generated contours by leveraging AI uncertainty and clinically-relevant features such as the proximity of the organs-at-risk to the tumor. Second, minimize the number of interactions needed to edit faulty delineations with redundancy-aware editing tools that provide the user a sense of predictability and control. We use a proof of concept that we validated with clinicians to demonstrate how current and upcoming AI capabilities support the workflow and how it would fit into clinical practice.@en

Deep learning models for organ contouring in radiotherapy are poised for clinical usage, but currently, there exist few tools for automated quality assessment (QA) of the predicted contours. Bayesian models and their associated uncertainty, can potentially automate the process of detecting inaccurate predictions. We investigate two Bayesian models for auto-contouring, DropOut and FlipOut, using a quantitative measure – expected calibration error (ECE) and a qualitative measure – region-based accuracy-vs-uncertainty (R-AvU) graphs. It is well understood that a model should have low ECE to be considered trustworthy. However, in a QA context, a model should also have high uncertainty in inaccurate regions and low uncertainty in accurate regions. Such behaviour could direct visual attention of expert users to potentially inaccurate regions, leading to a speed-up in the QA process. Using R-AvU graphs, we qualitatively compare the behaviour of different models in accurate and inaccurate regions. Experiments are conducted on the MICCAI2015 Head and Neck Segmentation Challenge and on the DeepMindTCIA CT dataset using three models: DropOut-DICE, Dropout-CE (Cross Entropy) and FlipOut-CE. Quantitative results show that DropOut-DICE has the highest ECE, while Dropout-CE and FlipOut-CE have the lowest ECE. To better understand the difference between DropOut-CE and FlipOut-CE, we use the R-AvU graph which shows that FlipOut-CE has better uncertainty coverage in inaccurate regions than DropOut-CE. Such a combination of quantitative and qualitative metrics explores a new approach that helps to select which model can be deployed as a QA tool in clinical settings.
@en

In automated driving, the user interface plays an essential role in guiding transitions between automated and manual driving. This literature review identified 25 studies that explicitly studied the effectiveness of user interfaces in automated driving. Our main selection criterion was how the user interface (UI) affected take-over performance in higher automation levels allowing drivers to take their eyes off the road (SAE3 and SAE4). We categorized user interface (UI) factors from an automated vehicle-related information perspective. Short take-over times are consistently associated with take-over requests (TORs) initiated by the auditory modality with high urgency levels. On the other hand, take-over requests directly displayed on non-driving-related task devices and augmented reality do not affect take-over time. Additional explanations of take-over situation, surrounding and vehicle information while driving, and take-over guiding information were found to improve situational awareness. Hence, we conclude that advanced user interfaces can enhance the safety and acceptance of automated driving. Most studies showed positive effects of advanced UI, but a number of studies showed no significant benefits, and a few studies showed negative effects of advanced UI, which may be associated with information overload. The occurrence of positive and negative results of similar UI concepts in different studies highlights the need for systematic UI testing across driving conditions and driver characteristics. Our findings propose future UI studies of automated vehicle focusing on trust calibration and enhancing situation awareness in various scenarios. @en

BACKGROUND: Former studies on work-related upper limb disorders (WRULD) within university education report substantial prevalence rates. In this study, developments in WRULD amongst students in the period 2004-2014 were investigated. Our findings can be a benchmark for future studies, in particular when there are major societal changes as in the case of the COVID-19 pandemic. OBJECTIVE: Differences in time (academic year), how long students have been studying (year of studying), relations with computer time and societal changes were points of interest. METHODS: 2254 students (average age 20.0 years) responded to a questionnaire on WRULD. Students experiencing complaints were further questioned about the severity of complaints and associated body locations. RESULTS: The average percentage of students experiencing complaints was 57%. The highest prevalence rates and severity scores were found in the first and last recorded academic years. The neck, shoulder, back and wrist were most often indicated. The prevalence of complaints raised from the 1st (49%) to the 4th (75%) year of studying. Two seriousness measures showed highest scores in the 5th/6th/7th year of studying. Relations were found between both the prevalence and seriousness of complaints with reported computer time. CONCLUSIONS: After an initial decreasing trend from the academic year 2006/2007 to 2010/2011 there was an increase in WRULD amongst students from 2010/2011 to 2013/2014. Limiting financial and study time factors may have played a role. Structural attention for WRULD prevention and risk factors seems to be effective in reducing prevalence and severity of WRULD. This seems to be even more necessary due to recent COVID-related changes in the students' lives.

@en

The authors explore the influence of the structure of a texture image on the perception of its color composition through a series of psychophysical studies. They estimate the color composition of a texture by extracting its dominant colors and the associated percentages. They then synthesize new textures with the same color composition but different geometric structural patterns. They conduct empirical studies in the form of two-alternative forced choice tests to determine the influence of two structural factors, pattern scale and shape, on the perceived amount of target color. The results of their studies indicate that (a) participants are able to consistently assess differences in color composition for textures of similar shape and scale, and (b) the perception of color composition is nonveridical. Pattern scale and shape have a strong influence on perceived color composition: the larger the scale, the higher the perceived amount of the target color, and the more elongated the shape, the lower the perceived amount of the target color. The authors also present a simple model that is consistent with the results of their empirical studies by accounting for the reduced visibility of the pixels near the color boundaries. In addition to a better understanding of human perception of color composition, their findings will contribute to the development of color texture similarity metrics.

@en

The development of autonomous driving systems cannot be achieved without the application of artificial intelligent (AI) systems. In Parasuraman et al (2000), It has been suggested that automation of systems shows parallels to the Human Information process. We will elaborate on this assumption in two steps. First, to integrate the Human Information Processing model (HIP) with the model of Rasmussen’s Human Performance model (1983). Second, to apply this integrated model on Technology Information Processing (TIP). We will show where they are similar and where they differ. Furthermore, we will argue that AI systems on a Knowledge Based level are not likely to occur in the near future@en