Trust and trustworthy artificial intelligence: A research agenda for AI in the environmental sciences.

Demands to manage the risks of artificial intelligence (AI) are growing. These demands and the government standards arising from them both call for trustworthy AI. In response, we adopt a convergent approach to review, evaluate, and synthesize research on the trust and trustworthiness of AI in the environmental sciences and propose a research agenda. Evidential and conceptual histories of research on trust and trustworthiness reveal persisting ambiguities and measurement shortcomings related to inconsistent attention to the contextual and social dependencies and dynamics of trust. Potentially underappreciated in the development of trustworthy AI for environmental sciences is the importance of engaging AI users and other stakeholders, which human-AI teaming perspectives on AI development similarly underscore. Co-development strategies may also help reconcile efforts to develop performance-based trustworthiness standards with dynamic and contextual notions of trust. We illustrate the importance of these themes with applied examples and show how insights from research on trust and the communication of risk and uncertainty can help advance the understanding of trust and trustworthiness of AI in the environmental sciences.

Lessons from the Glass Cockpit: Innovation in Alarm Systems to Support Cognitive Work.

Nurses working in the hospital setting increasingly have become overburdened by managing alarms that, in many cases, provide low information value regarding patient health. The current trend, aided by disposable, wearable technologies, is to promote patient monitoring that does not require entering a patient's room. The development of telemetry alarms and middleware escalation devices adds to the continued growth of auditory, visual, and haptic alarms to the hospital environment but can fail to provide a more complete understanding of patient health. As we begin to innovate to both address alarm overload and improve patient management, perhaps using fundamentally different integration architectures, lessons from the aviation flight deck are worth considering. Commercial jet transport systems and their alarms have evolved slowly over many decades and have developed integration methods that account for operational context, provide multiple response protocol levels, and present a more integrated view of the airplane system state. We articulate three alarm system objectives: (1) supporting hazard management, (2) establishing context, and (3) supporting alarm prioritization. More generally, we present the case that alarm design in aviation can spur directions for innovation for telemetry monitoring systems in hospitals.

Ten Best Practices for Improving Emergency Medicine Provider-Nurse Communication.

BACKGROUND: The current state of scientific knowledge regarding communication between emergency medicine (EM) providers indicates that communication is critical to safe and effective patient care. OBJECTIVES: In this study, we identified communication needs of EM nurses and physicians; in particular, what information should be conveyed, when, how, and to whom. METHODS: Five semi-structured focus groups and one interview were conducted with nine nurses, eight attending physicians, and four residents. Questions addressed how EM personnel use and share information about patients and clinical work, what information tends to be exchanged, and what additional information would be helpful to share. Sessions were audio recorded. Transcripts were generated and analyzed using a concept mapping approach (a visual qualitative analysis technique to represent and convey synthesized knowledge). RESULTS: Eleven concept maps were produced summarizing: information physicians needed from nurses and vice versa; methods of communication that could be utilized; barriers or obstacles to effective communication; strategies to enhance or ensure effective communication; and environmental or situational factors that impact communication. CONCLUSIONS: Our main finding of this research is that communication ensures shared awareness of patient health status, the care plan, status of plan steps and orders, and, especially, any critical changes or "surprises" regarding the health of a patient. Additionally, the research identified shared information needs; communication methods, strategies and barriers; and factors affecting successful communication, and has implications for both system and training design. Key implications for emergency nursing practice from this research are distilled in 10 'best practice' strategies for improving EM nurse-physician communication.

Cognitive engineering and health informatics: Applications and intersections.

Cognitive engineering is an applied field with roots in both cognitive science and engineering that has been used to support design of information displays, decision support, human-automation interaction, and training in numerous high risk domains ranging from nuclear power plant control to transportation and defense systems. Cognitive engineering provides a set of structured, analytic methods for data collection and analysis that intersect with and complement methods of Cognitive Informatics. These methods support discovery of aspects of the work that make performance challenging, as well as the knowledge, skills, and strategies that experts use to meet those challenges. Importantly, cognitive engineering methods provide novel representations that highlight the inherent complexities of the work domain and traceable links between the results of cognitive analyses and actionable design requirements. This article provides an overview of relevant cognitive engineering methods, and illustrates how they have been applied to the design of health information technology (HIT) systems. Additionally, although cognitive engineering methods have been applied in the design of user-centered informatics systems, methods drawn from informatics are not typically incorporated into a cognitive engineering analysis. This article presents a discussion regarding ways in which data-rich methods can inform cognitive engineering.

Defining the methodological challenges and opportunities for an effective science of sociotechnical systems and safety.

An important part of the application of sociotechnical systems theory (STS) is the development of methods, tools and techniques to assess human factors and ergonomics workplace requirements. We focus in this paper on describing and evaluating current STS methods for workplace safety, as well as outlining a set of six case studies covering the application of these methods to a range of safety contexts. We also describe an evaluation of the methods in terms of ratings of their ability to address a set of theoretical and practical questions (e.g. the degree to which methods capture static/dynamic aspects of tasks and interactions between system levels). The outcomes from the evaluation highlight a set of gaps relating to the coverage and applicability of current methods for STS and safety (e.g. coverage of external influences on system functioning; method usability). The final sections of the paper describe a set of future challenges, as well as some practical suggestions for tackling these. PRACTITIONER SUMMARY: We provide an up-to-date review of STS methods, a set of case studies illustrating their use and an evaluation of their strengths and weaknesses. The paper concludes with a 'roadmap' for future work.

Protecting patients from an unsafe system: the etiology and recovery of intraoperative deviations in care.

OBJECTIVE: To understand the etiology and resolution of unanticipated events in the operating room (OR). BACKGROUND: The majority of surgical adverse events occur intraoperatively. The OR represents a complex, high-risk system. The influence of different human, team, and organizational/environmental factors on safety and performance is unknown. METHODS: We video-recorded and transcribed 10 high-acuity operations, representing 43.7 hours of patient care. Deviations, defined as delays and/or episodes of decreased patient safety, were identified by majority consensus of a multidisciplinary team. Factors that contributed to each event and/or mitigated its impact were determined and attributed to the patient, providers, or environment/organization. RESULTS: Thirty-three deviations (10 delays, 17 safety compromises, 6 both) occurred--with a mean of 1 every 79.4 minutes. These deviations were multifactorial (mean 3.1 factors). Problems with communication and organizational structure appeared repeatedly at the root of both types of deviations. Delays tended to be resolved with vigilance, communication, coordination, and cooperation, while mediation of safety compromises was most frequently accomplished with vigilance, leadership, communication, and/or coordination. The organization/environment was not found to play a direct role in compensation. CONCLUSIONS: Unanticipated events are common in the OR. Deviations result from poor organizational/environmental design and suboptimal team dynamics, with caregivers compensating to avoid patient harm. Although recognized in other high-risk domains, such human resilience has not yet been described in surgery and has major implications for the design of safety interventions.

Deconstructing intraoperative communication failures.

BACKGROUND: Communication failure is a common contributor to adverse events. We sought to characterize communication failures during complex operations. METHODS: We video recorded and transcribed six complex operations, representing 22 h of patient care. For each communication event, we determined the participants and the content discussed. Failures were classified into four types: audience (key individuals missing), purpose (issue nonresolution), content (insufficient/inaccurate information), and/or occasion (futile timing). We added a systems category to reflect communication occurring at the organizational level. The impact of each identified failure was described. RESULTS: We observed communication failures in every case (mean 29, median 28, range 13-48), at a rate of one every 8 min. Cross-disciplinary exchanges resulted in failure nearly twice as often as intradisciplinary ones. Discussions about or mandated by hospital policy (20%), personnel (18%), or other patient care (17%) were most error prone. Audience and purpose each accounted for >40% of failures. A substantial proportion (26%) reflected flawed systems for communication, particularly those for disseminating policy (29% of system failures), coordinating personnel (27%), and conveying the procedure planned (27%) or the equipment needed (24%). In 81% of failures, inefficiency (extraneous discussion and/or work) resulted. Resource waste (19%) and work-arounds (13%) also were frequently seen. CONCLUSIONS: During complex operations, communication failures occur frequently and lead to inefficiency. Prevention may be achieved by improving synchronous, cross-disciplinary communication. The rate of failure during discussions about/mandated by policy highlights the need for carefully designed standardized interventions. System-level support for asynchronous perioperative communication may streamline operating room coordination and preparation efforts.

Modeling patient-related workload in the emergency department using electronic health record data.

INTRODUCTION: Understanding and managing clinician workload is important for clinician (nurses, physicians and advanced practice providers) occupational health as well as patient safety. Efforts have been made to develop strategies for managing clinician workload by improving patient assignment. The goal of the current study is to use electronic health record (EHR) data to predict the amount of work that individual patients contribute to clinician workload (patient-related workload). METHODS: One month of EHR data was retrieved from an emergency department (ED). A list of workload indicators and five potential workload proxies were extracted from the data. Linear regression and four machine learning classification algorithms were utilized to model the relationship between the indicators and the proxies. RESULTS: Linear regression proved that the indicators explained a substantial amount of variance of the proxies (four out of five proxies were modeled with R2 > 0.80). Classification algorithms also showed success in classifying a patient as having high or low task demand based on data from early in the ED visit (e.g. 80 % accurate binary classification with data from the first hour). CONCLUSION: The main contribution of this study is demonstrating the potential of using EHR data to predict patient-related workload automatically in the ED. The predicted workload can potentially help in managing clinician workload by supporting decisions around the assignment of new patients to providers. Future work should focus on identifying the relationship between workload proxies and actual workload, as well as improving prediction performance of regression and multi-class classification.

Design and Evaluation of an Integrated, Patient-Focused Electronic Health Record Display for Emergency Medicine.

BACKGROUND: Hospital emergency departments (EDs) are dynamic environments, involving coordination and shared decision making by staff who care for multiple patients simultaneously. While computerized information systems have been widely adopted in such clinical environments, serious issues have been raised related to their usability and effectiveness. In particular, there is a need to support clinicians to communicate and maintain awareness of a patient's health status, and progress through the ED plan of care. OBJECTIVE: This study used work-centered usability methods to evaluate an integrated patient-focused status display designed to support ED clinicians' communication and situation awareness regarding a patient's health status and progress through their ED plan of care. The display design was informed by previous studies we conducted examining the information and cognitive support requirements of ED providers and nurses. METHODS: ED nurse and provider participants were presented various scenarios requiring patient-prioritization and care-planning tasks to be performed using the prototype display. Participants rated the display in terms of its cognitive support, usability, and usefulness. Participants' performance on the various tasks, and their feedback on the display design and utility, was analyzed. RESULTS: Participants provided ratings for usability and usefulness for the display sections using a work-centered usability questionnaire-mean scores for nurses and providers were 7.56 and 6.6 (1 being lowest and 9 being highest), respectively. General usability scores, based on the System Usability Scale tool, were rated as acceptable or marginally acceptable. Similarly, participants also rated the display highly in terms of support for specific cognitive objectives. CONCLUSION: A novel patient-focused status display for emergency medicine was evaluated via a simulation-based study in terms of work-centered usability and usefulness. Participants' subjective ratings of usability, usefulness, and support for cognitive objectives were encouraging. These findings, including participants' qualitative feedback, provided insights for improving the design of the display.

Impact of clinical reminder redesign on learnability, efficiency, usability, and workload for ambulatory clinic nurses.

OBJECTIVE: Computerized clinical reminders (CRs) were designed to reduce clinicians' reliance on their memory and to present evidence-based guidelines at point of care. However, the literature indicates that CR adoption and effectiveness has been variable. We examined the impact of four design modifications to CR software on learnability, efficiency, usability, and workload for intake nursing personnel in an outpatient clinic setting. These modifications were included in a redesign primarily to address barriers to effective CR use identified during a previous field study. DESIGN: In a simulation experiment, 16 nurses used prototypes of the current and redesigned system in a within-subject comparison for five simulated patient encounters. Prior to the experimental session, participants completed an exploration session, where "learnability" of the current and redesigned systems was assessed. MEASUREMENTS: Time, performance, and survey data were analyzed in conjunction with semi-structured debrief interview data. RESULTS: The redesign was found to significantly increase learnability for first-time users as measured by time to complete the first CR, efficiency as measured by task completion time for two of five patient scenarios, usability as determined by all three groupings of questions taken from a commonly used survey instrument, and two of six workload subscales of the NASA Task Load Index (TLX) survey: mental workload and frustration. CONCLUSION: Modest design modifications to existing CR software positively impacted variables that likely would increase the willingness for first-time nursing personnel to adopt and consistently use CRs.

Communication in the Electronic Age: an Analysis of Face-to-Face Physician-Nurse Communication in the Emergency Department.

We describe the patterns and content of nurse to physician verbal conversations in three emergency departments (EDs) with electronic health records. Emergency medicine physicians and nurses were observed for 2 h periods. Researchers used paper notes to document the characteristics (e.g., partners involved, location of communication, who initiated communication) and content of nurse to physician conversations. Eighteen emergency nurses and physicians (nine each) were observed for a total of 36 h. Two hundred and fifty-five unique communication events were recorded across three emergency departments spread evenly across day, evening, and night shifts. A qualitative analysis of communication event content revealed 5 types of communication and 13 content themes. Content themes covered a broad range of topics including exchange of patient health information, management of the ED, and coordination of orders. Physician participants experienced significantly more communication events than nurse participants, while nurses initiated significantly more communication events than physicians. Most of the communication events occurred at the physician workstation followed by patient treatment areas. This study describes nature of verbal nurse to physician communication in the ED. Direct communication is still used to communicate important information, such as information about patients' status, in EDs with established electronic health records. Our results provide an overview of information exchanged in the ED which can serve as a basis for designing improved information support systems.

A prospective study of patient safety in the operating room.

BACKGROUND: To better understand the operating room as a system and to identify system features that influence patient safety, we performed an analysis of operating room patient care using a prospective observational technique. METHODS: A multidisciplinary team comprised of human factors experts and surgeons conducted prospective observations of 10 complex general surgery cases in an academic hospital. Minute-to-minute observations were recorded in the field, and later coded and analyzed. A qualitative analysis first identified major system features that influenced team performance and patient safety. A quantitative analysis of factors related to these systems features followed. In addition, safety-compromising events were identified and analyzed for contributing and compensatory factors. RESULTS: Problems in communication and information flow, and workload and competing tasks were found to have measurable negative impact on team performance and patient safety in all 10 cases. In particular, the counting protocol was found to significantly compromise case progression and patient safety. We identified 11 events that potentially compromised patient safety, allowing us to identify recurring factors that contributed to or mitigated the overall effect on the patient's outcome. CONCLUSIONS: This study demonstrates the role of prospective observational methods in exposing critical system features that influence patient safety and that can be the targets for patient safety initiatives. Communication breakdown and information loss, as well as increased workload and competing tasks, pose the greatest threats to patient safety in the operating room.

Making the operating room of the future safer.

There is an increasing demand for interventions to improve patient safety, but there is limited data to guide such reform. In particular, because much of the existing research is outcome-driven, we have a limited understanding of the factors and process variations that influence safety in the operating room. In this article, we start with an overview of safety terminology, suggesting a model that emphasizes "safety" rather than "error" and that can encompass the spectrum of events occurring in the operating room. Next, we provide an introduction to techniques that can be used to understand safety at the point of care and we review the data that exists relating such studies to improved outcomes. Future work in this area will need to prospectively study the processes and factors that impact patient safety and vulnerability in the operating room.

Usability in Health IT: Beyond Compliance to Meaningful Design and Assessment.

The relatively rapid transition from a paper-based system to a digital system in healthcare has not always employed a sophisticated integration of usability concepts. Yet usability is critical to safety and to effectiveness of the electronic health record, and regulators and policy makers have been increasingly focused on this area. This panel will provide a variety of perspectives on this important issue, ranging from a description of the problem based on current vendor usability practices; recommendations regarding domain content rich usability processes including use cases, assessments, and scenarios; and the extension of usability assessments and design improvements to post-system implementation.

Surgeons' Leadership Styles and Team Behavior in the Operating Room.

BACKGROUND: The importance of leadership is recognized in surgery, but the specific impact of leadership style on team behavior is not well understood. In other industries, leadership is a well-characterized construct. One dominant theory proposes that transactional (task-focused) leaders achieve minimum standards and transformational (team-oriented) leaders inspire performance beyond expectations. STUDY DESIGN: We videorecorded 5 surgeons performing complex operations. Each surgeon was scored on the Multifactor Leadership Questionnaire, a validated method for scoring transformational and transactional leadership style, by an organizational psychologist and a surgeon researcher. Independent coders assessed surgeons' leadership behaviors according to the Surgical Leadership Inventory and team behaviors (information sharing, cooperative, and voice behaviors). All coders were blinded. Leadership style (Multifactor Leadership Questionnaire) was correlated with surgeon behavior (Surgical Leadership Inventory) and team behavior using Poisson regression, controlling for time and the total number of behaviors, respectively. RESULTS: All surgeons scored similarly on transactional leadership (range 2.38 to 2.69), but varied more widely on transformational leadership (range 1.98 to 3.60). Each 1-point increase in transformational score corresponded to 3 times more information-sharing behaviors (p < 0.0001) and 5.4 times more voice behaviors (p = 0.0005) among the team. With each 1-point increase in transformational score, leaders displayed 10 times more supportive behaviors (p < 0.0001) and displayed poor behaviors 12.5 times less frequently (p < 0.0001). Excerpts of representative dialogue are included for illustration. CONCLUSIONS: We provide a framework for evaluating surgeons' leadership and its impact on team performance in the operating room. As in other fields, our data suggest that transformational leadership is associated with improved team behavior. Surgeon leadership development, therefore, has the potential to improve the efficiency and safety of operative care.

War stories: a qualitative analysis of narrative teaching strategies in the operating room.

BACKGROUND: "War stories" are commonplace in surgical education, yet little is known about their purpose, construct, or use in the education of trainees. METHODS: Ten complex operations were videotaped and audiotaped. Narrative stories were analyzed using grounded theory to identify emergent themes in both the types of stories being told and the teaching objectives they illustrated. RESULTS: Twenty-four stories were identified in 9 of the 10 cases (mean, 2.4/case). They were brief (mean, 58 seconds), illustrative of multiple teaching points (mean, 1.5/story), and appeared throughout the operations. Anchored in personal experience, these stories taught both clinical (eg, operative technique, decision making, error identification) and programmatic (eg, resource management, professionalism) topics. CONCLUSIONS: Narrative stories are used frequently and intuitively by physicians to emphasize a variety of intraoperative teaching points. They socialize trainees in the culture of surgery and may represent an underrecognized approach to teaching the core competencies. More understanding is needed to maximize their potential.

Uncovering the requirements of cognitive work.

OBJECTIVE: In this article, the author provides an overview of cognitive analysis methods and how they can be used to inform system analysis and design. BACKGROUND: Human factors has seen a shift toward modeling and support of cognitively intensive work (e.g., military command and control, medical planning and decision making, supervisory control of automated systems). Cognitive task analysis and cognitive work analysis methods extend traditional task analysis techniques to uncover the knowledge and thought processes that underlie performance in cognitively complex settings. METHODS: The author reviews the multidisciplinary roots of cognitive analysis and the variety of cognitive task analysis and cognitive work analysis methods that have emerged. RESULTS: Cognitive analysis methods have been used successfully to guide system design, as well as development of function allocation, team structure, and training, so as to enhance performance and reduce the potential for error. CONCLUSIONS: A comprehensive characterization of cognitive work requires two mutually informing analyses: (a) examination of domain characteristics and constraints that define cognitive requirements and challenges and (b) examination of practitioner knowledge and strategies that underlie both expert and error-vulnerable performance. A variety of specific methods can be adapted to achieve these aims within the pragmatic constraints of particular projects. APPLICATION: Cognitive analysis methods can be used effectively to anticipate cognitive performance problems and specify ways to improve individual and team cognitive performance (be it through new forms of training, user interfaces, or decision aids).

Lessons from a comparison of work domain models: representational choices and their implications.

As methods in cognitive work analysis become more widely applied, questions regarding the impact of modeling choices and similarities in modeling efforts across projects and domains are increasingly relevant. However, no explicit comparison of models of similar systems has been reported. This paper compares independently developed work domain analysis (WDA) models of two command and control environments. Similarities in model content and the types of nodes included provide evidence that WDA techniques can capture fundamental elements regarding purposes and constraints. These points of agreement provide a common starting point for developing work domain representations of military command and control systems. The comparison also revealed differences between the models. Although differences in content reflected differences in scope of coverage and level of detail, other differences corresponded to more fundamental choices in modeling approach. These included the treatment of sensors, level of integration in the model, and representation of particular abstract constraints. Examination of these more fundamental differences pointed to important degrees of freedom in how to represent a WDA and clarified the implications of these modeling choices for guiding design. Actual or potential applications of this research include aiding analysts in making work domain modeling choices as well as producing work domain models of command and control environments.

Handoff strategies in settings with high consequences for failure: lessons for health care operations.

OBJECTIVE: To describe strategies employed during handoffs in four settings with high consequences for failure. ANALYSIS: of observational data for evidence of use of 21 handoff strategies. SETTING: NASA Johnson Space Center in Texas, nuclear power generation plants in Canada, a railroad dispatch center in the United States, and an ambulance dispatch center in Toronto. MAIN MEASURE: Evidence of 21 handoff strategies from observations and interviews. RESULTS: Nineteen of 21 strategies were used in at least one domain, on at least an 'as needed' basis. CONCLUSIONS: An understanding of how handoffs are conducted in settings with high consequences for failure can jumpstart endeavors to modify handoffs to improve patient safety.