Abstract: Low Power Wide Area Network (LPWAN) technologies are typically promoted for Internet-of-Things (IoT) applications, but are also of interest for emergency communications systems when regular fixed and mobile networks break down. Although LoRaWAN is a frequently used representative here, there are sometimes large differences between the proposed range and the results of some practical evaluations. Since previous work has focused on urban environments or has conducted simulations, this work aims to gather concrete knowledge on the transmission characteristics in rural environments. Extensive field studies with varying geographic conditions and comparative tests in urban environments were performed using two different hardware implementations. Overall, it was found that the collected values in rural areas are significantly lower than the theoretical values. Nevertheless, the results certify that LoRaWAN technology has a high range that cannot be achieved with other common technologies for emergency communications.

Abstract: Search and rescue (SAR) teams are the first to respond to emergencies. This could include finding lost hikers, shoring buildings, or aiding people post-disaster. SAR combines orienteering, engineering, field medicine, and communication. Technology use in SAR has been changing with the proliferation of information communication technologies; so, we ask, how are established and emerging technologies used in SAR? Understanding how responders are adopting and adapting these technologies during SAR missions can inform future design and improve outcomes for SAR teams. We interviewed SAR volunteers to contextualize their experiences with technology and triangulated with additional questionnaire data. We discuss how technology use in SAR requires an intersection of expert knowledge and creative problem solving to overcome challenges in the field. This research contributes an understanding of the constraints on and implications for future SAR technologies and SAR operators’ creativity in emergent situations.

Abstract: The course of large-scale incidents as well as disasters can reveal weaknesses in command and control (C2) systems, which make adjustments necessary. Also, new technologies may require C2-systems to be adapted to achieve their full potential for improving incident command. This paper deals with an approach to enable the comparison and evaluation of different C2-systems or their adaptations in order to find the best possible customizations for C2-systems. To this purpose, systems theory is used to unify the approaches of different research disciplines. Within the C2-system boundaries, distinctions were made to represent three different levels of evaluation: “Physical Characteristics”, “Structures and Processes” as well as “C2-system-effectiveness”. During the implementation of the evaluation methods from the different research disciplines into the systems theory approach, it became apparent that the comprehensive approach is desirable, but that broad knowledge and expertise is necessary, especially at the highest evaluation level “C2-system-effectiveness”.

Abstract: Understanding the COVID-19 pandemic impacts on low-income households can inform social services about the needs of vulnerable communities. Some recent works have studied such impacts through social media content analysis, and supervised machine learning models have been proposed to automatically classify COVID-19 tweets into different categories, such as income and economy impacts, social inequality and justice issues, etc. In this paper, we propose semi-supervised learning models based on BERT with Self-Training and Knowledge Distillation for identifying COVID-19 tweets relevant to low-income households by leveraging readily available unlabeled data in addition to limited amounts of labeled data. Furthermore, we explore ChatGPT’s potential for annotating COVID-19 data and the performance of fine-tuned GPT-3 models. Our semi-supervised BERT model with Knowledge Distillation showed improvements compared to a supervised baseline model, while zero-shot ChatGPT showed good potential as a tool for annotating crisis data. However, our study suggests that the cost of fine-tuning large and expensive GPT-3 models may not be worth for some tasks.

Abstract: In times of emergency, crisis response agencies need to quickly and accurately assess the situation on the ground in order to deploy relevant services and resources. However, authorities often have to make decisions based on limited information, as data on affected regions can be scarce until local response services can provide first-hand reports. Fortunately, the widespread availability of smartphones with high-quality cameras has made citizen journalism through social media a valuable source of information for crisis responders. However, analyzing the large volume of images posted by citizens requires more time and effort than is typically available. To address this issue, this paper proposes the use of state-of-the-art deep neural models for automatic image classification/tagging, specifically by adapting transformer-based architectures for crisis image classification (CrisisViT). We leverage the new Incidents1M crisis image dataset to develop a range of new transformer-based image classification models. Through experimentation over the standard Crisis image benchmark dataset, we demonstrate that the CrisisViT models significantly outperform previous approaches in emergency type, image relevance, humanitarian category, and damage severity classification. Additionally, we show that the new Incidents1M dataset can further augment the CrisisViT models resulting in an additional 1.25% absolute accuracy gain.

Abstract: Supervision of technical dives is particularly important in emergency and disaster response operations to ensure the safety of divers in unexplored locations with uncertain conditions. Diver monitoring relies primarily on voice communication and a video stream that gives the operator a first-person view of the diver. However, in many cases underwater visibility can drop to just a few centimeters, leaving the diver only able to feel his way with his hands and the operator depended only on voice communication, making it very difficult for both of them to identify upcoming hazards. In the DeeperSense research project, we are attempting to reduce the limitations caused by poor underwater visibility by using a sonar in combination with an AI-based algorithm designed to translate the sonar signal into a visual image that is independent of the turbidity of the water and gives an overview of the situation where the eye can no longer see anything. Laboratory results show that visual information can be recovered from sonar data.

Abstract: Assessing the effectiveness of crisis response is key to improving preparedness and adapting policies. One method for response evaluation is reviewing actual response activities and interactions. Response reports are often available in the form of natural language text data. Analyzing a large number of such reports requires automated or semi automated solutions. To improve the trustworthiness of methods for this purpose, we empirically validate the reliability of three relation extraction methods that we used to construct interorganizational collaboration networks by comparing them against human-annotated ground truth (crisis-specific situational reports and tweets). For entity extraction, we find that using a combination of two off-the-shelf methods (FlairNLP and SpaCy) is optimal for situational reports data and one method (SpaCy) for tweets data. For relation extraction, we find that a heuristics-based model that we built by leveraging word co-occurrence and deep and shallow syntax as features and training it on domain-specific text data outperforms two state-of-the-art relation extraction models (Stanford OpenIE and OneIE) that were pre-trained on general domain data. We also find that situational reports, on average, contain less entities and relations than tweets, but the extracted networks are more closely related to collaboration activities mentioned in the ground truth. As it is widely known that general domain tools might need adjustment to perform accurately in specific domains, we did not expect the tested off-the-shelf tools to perform highly accurately. Our point is to rather identify what accuracy one could reasonably expect when leveraging available resources as-is for domain specific work (in this case, crisis informatics), what errors (in terms of false positives and false negatives) to expect, and how to account for that.

Abstract: Between 2018 and 2021, the Incident Streams track (TREC-IS) developed standard approaches for classifying information types and criticality of tweets during crises. While successful in producing substantial collections of labeled data, TREC-IS as a data challenge had several limitations: It only evaluated information at type-level rather than what was reported; it only used Twitter data; and it lacked measures of redundancy in system output. This paper introduces Crisis Facts and Cross-Stream Temporal Summarization (CrisisFACTS), a new data challenge piloted in 2022 and developed to address these limitations. The CrisisFACTS framework recasts TREC-IS into an event-summarization task using multiple disaster-relevant data streams and a new fact-based evaluation scheme, allowing the community to assess state-of-the-art methods for summarizing disaster events Results from CrisisFACTS in 2022 include a new test-collection comprising human-generated disaster summaries along with multi-platform datasets of social media, crisis reports and news coverage for major crisis events.

Abstract: The chances of survival of a victim of cardiac arrest or stroke decrease considerably without rapid intervention. Bystanders, the first people able to intervene, are however few to act. This ignorance of the seriousness signs of the pathologies, and of the importance of acting, combined with a feeling of incapacity to perform emergency gestures, are the main reasons for this low level of action. The absence of knowledge originates both from a lack of training and from forgetting training courses. To overcome this problem, some trainings propose high fidelity simulation devices associated with a debriefing to create a strong emotional impact leading to a stronger memory impact. To assess the impact of this type of simulation, we set up a Living Lab including a high fidelity simulation of emergency situations (with citizens, dispatchers, first responders and paramedics), a debriefing and a method to create a “trace of activity”, still aiming at generating a higher memory impact. To measure the effects of the Living Lab, we analyzed the emotional impacts evoked by the participants, categorized the learning and finally studied the creation of the activity trace. The results show that the Living-Lab elicits emotions (for the cardiac arrest scenario) and projections (for the stroke scenario) and can therefore potentially improve the retention. The learnings were of several natures: individual and practical learnings on emergency management, learnings on collaboration within the chain of survival and theoretical learnings. Analysis of the retained learning after 2 to 8 months is in progress, therefore no results are available yet.

Abstract: Emergency response services face massive pressure during global crises, such as COVID-19. The food supply logistics sector is one of the pressures that impacted the emergency response services, due to crisis restrictions. A regulatory framework to deploy autonomous vehicles, in any nominated country, has been presented to boost the food supply logistics as an emergency response to critical situations to serve isolated areas. This framework resulted in three steps to deploy AVs in the nominated country, which are evaluating their legislation, modifying their existing regulations accordingly, and ensuring the full deployment of the innovative technology. This is done by minimising person-to-person contact during the transportation and distribution phase. In conclusion, fully autonomous vehicles can help lift the pressure from the emergency response teams in the food supply transportation and distribution phase to meet the basic living requirements for human needs during global crises.

Abstract: Out-of-hospital cardiac arrest (OHCA) and choking are two emergencies where the rapid action of a bystander can increase the victim's chances of survival. Few bystanders act because they are not aware of their role as the first link in the chain of survival. Working on collaboration among a local chain of survival and using applications to improve communication and provide tutorials of actions to perform can be used to overcome this issue. We investigate these elements in the context of the Geneva Chain of Survival using simulations. The results show that an optimal collaboration means a lead’s handover between the intervening parties. Collaboration can be degraded by problems of communication, panic , and confusion. Applications constitute a valuable addition to enhance the dispatcher's awareness and to help guide the CPR while not extending the intervention time. Finally, the debriefing that follows enables the acquisition of competencies through experiential learning that relies on emotions.

Abstract: The local government’s continuous support is critical for the well-being of a community during disaster events. E-Government systems that establish and maintain ongoing connections with the community thus play a vital role in supporting crisis response and recovery. Such systems’ ability to adapt to the crisis circumstances and to address emergent needs helps them continue their fundamental functions during disasters. Considering various services might require different amounts and types of resources, prioritization strategies are helpful in determining the processing order of requests. This paper discusses the role of prioritizing services within an e-Government system, to better understand how such a system can be managed to best utilize available resources. The study examines how a well-functioning e-Government system, the Orange County, Florida 311 non-emergency service system, responded to the COVID-19 pandemic and how the changes in service operations requirements can affect service provision, specifically with respect to assigning or re-assigning priority levels.