Models of Cognitive Work Analysis in Neurocritical Care and Perspectives on Expertise– An Ecological Interface Design Approach

Abstract

Neurocritical care is a complex and data-rich environment in which timely decisions are of utmost importance. Small changes in the neurophysiologic states may imply neurological deterioration that can be fatal if not treated timely and appropriately. A multidisciplinary team of clinicians takes care of the patients with varying expertise levels. Expert physicians in this domain often try to act proactively and prevent further injury (such as secondary brain injury often visible in traumatic brain injury patients) and manage concurrently, assessing patient data represented through various sources. Novice intensive care physicians act rather reactively to specific events and have a sequential approach to management. The scattered data sources across the unit and the lack of centralized views on the patient data make it even more challenging for physicians, in general, to understand the happenings and support the patient. Collecting and identifying all data through various locations and systems can cause a high cognitive workload and is not optimally laid out through meaningful representations on different device interfaces. Further, such interfaces (e.g., the bedside physiologic monitor) lack the incorporation of neurocritical-relevant concept visualizations that would help novices better analyze and understand the context and significant relationships among variables. This is especially important because the complexity of neurocritical care is high, immediate, and proactive treatment is desired. It usually takes much time for novices to learn the specifics of assessing neurological deterioration in a critical context and being aware of the situation to consider the trajectory and required medical support for the patient. Inappropriate actions and long decision-making time can result in unwanted health deteriorations or even fatalities. Bedside physiologic monitors play a significant role in assessing the vital signs of patients; however, in neurocritical care, there are additional variables, such as Intracranial Pressure (ICP), that are significant biomarkers for neurological conditions and need to be assessed in context with other variables. In this work, we have focused on this specific variable as it is a key variable showing the health state of the brain and is interconnected with various other physiologic variables, which requires a detailed examination to assess neurocritical care patients’ conditions. For novice neurocritical care physicians, it is a new variable and challenging context to learn how to deal with this variable as it was often displayed numerically and as a waveform on monitors. Assessing this variable in context needs to be trained to develop an understanding of (dys)functional autoregulation in the brain. Still, interfaces don’t have much flexibility in representing this variable other than in a numeric or waveform pattern. Monitors thus can benefit from new representations of ICP within its context. This work proposes an ecological interface design (EID) approach to represent ICP together with other variables on vital signs monitor to be used in neurocritical care. This work not only shows a new design for such monitor interfaces but also integrates the additional perspective on supporting novice physicians to develop more in the direction of experts to support them in the best way possible. Differences in the context of neurocritical care among physicians are explored in this work, as they are the main decision-makers in the unit. Once the understanding of differences between novices and experts can be disseminated, the steps that are required to develop expertise can be further examined and supported through various ways. Overall, the research objective of this work is to explore how expertise development can be supported through the ecological interface design approach of bedside physiologic monitors in the context of neurocritical care.

Methods and Modeling: As a starting point, observations in a neurocritical care unit are summarized and modeled through the Cognitive Work Analysis (CWA) framework using the Work Domain Analysis (WDA) and Control Task Analysis (ConTa). This framework is selected to identify various elements of the neurocritical care context as it provides in-depth insights into complex sociotechnical areas and their constraints. Further, it lays a basis for the design of an ecological interface that enables the end-user to become an adaptive problem solver in any situation. The CWA models thus provide the basis for identifying key relationships and aspects that would be important to display on such interfaces.

In the following step, interviews were conducted with critical care experts and novices to further dive into their perspectives on expertise development, challenges, and potential ways of addressing these and supporting novices. These insights were relevant for adding and discussing aspects represented on the CWA models and exploring potential expertise-relevant measures that could be used for assessing the development of expertise through interface design.

Identifying both CWA and interview findings, static visualizations were developed and discussed with both novice and expert critical care physicians. Their feedback was incorporated into the new designs of the visualizations and integrated into a prototype (i.e., the ecological interface). This interface was compared to a standard interface used in the neurocritical care unit within a usability study. Two neurocritical care scenarios were presented to the participants in the control and experimental groups while being asked to think out loud about assessing and treating such patient cases based on the interfaces and data shown.

Results: The CWA models showed different dimensions of the work domain and how the physical variables are interconnected to provide and support patient recovery. The decision ladders mapped on a common neurocritical care scenario showed differences in expert and novice physicians’ approaches to assessing, evaluating, and acting on certain signs related to the health states of a patient. While novices cognitively go through each step and might miss out on case-relevant information or may come up with misinterpretations even, expert physicians take shortcuts as they have built certain mental models already based on their vast knowledge and experience with similar cases. Through the interviews and discussions with physicians, we further noticed that novice critical care physicians usually have a numerical and threshold approach. At the same time, experts expect them to develop an approach that incorporates understanding the significance of the waveform, trend, and individualize patient care depending on various patient characteristics. To further develop their mental model, they must step back from the “one size fits all” mindset and explore the optimal thresholds for each patient individually. To progress, they further need to develop autonomy and communication skills with various stakeholders, deepen their knowledge and familiarity with tools, perform adequately, and reflect on their actions. The interviews also revealed various aspects of expertise often tracked by the expert physicians training the novices. It was noted that experts often have a great amount of experience, knowledge, skills, training, work where there is no evidence, have excellent performance and specific personal traits. The inputs from both the CWA models and the discussions with the physicians provided initial design ideas that could be further developed to support aspects of the novices' expertise development process in the interface design context. Concepts that are relevant for neurocritical care monitoring and associated challenges have thus been discussed. Both expert and novice critical care physicians provided feedback on the initial visualizations, which helped to iterate the visualizations further and were then incorporated into a prototype interface (i.e., the ecological interface). A standard interface was also developed as a prototype to compare how participants in both groups could be compared. The intention was to show one potential ecological interface for neurocritical care that would specifically support the novice physicians to develop more characteristics of experts identified in the proposed studies. Novices using the ecological interface (experimental group) showed fewer errors when presented with two different neurocritical care scenarios, compared to the control group using the standard interface. At the same time, the experimental group’s reasoning was more focused and included more investigation of the cause for ICP elevations and patterns compared to the group using a standard interface. Reflections on the ecological interface further showed that participants started thinking critically about their strategies and helped them state the reasons and limitations of their actions demonstrated during the scenario presentations. The confidence in handing over the patient to the next team has been rated on a numeric scale showing that the group using the standard interface rated slightly higher on average than the experimental group. Also, the average subjective performance ratings were lower for the ecological interface than the standard interface. It is noteworthy though, that the difference between both groups on confidence and performance ratings wasn’t too far away from another. However, the usability of the ecological interface was perceived as useful, easy to use, and captured key neurocritical care concepts relevant to further develop expertise in the field. The ecological interface was perceived as an improvement to current interfaces. Recommendations on future interface design are shared and can help future research to further improve interface design.

Conclusion: This research work consisted of multiple phases uniting ethnographic observations, discussions with clinicians (especially novice and expert physicians in neurocritical care), the development stage of neurocritical care-relevant concepts that are lacking in current bedside physiologic monitor interface designs, an iterative design process of potential visualizations to be displayed on such interfaces, and the usability evaluation of the visualizations in context (with novice neurocritical care physicians). Although the findings in the usability evaluation haven’t shown exact confirmation of all expertise development hypotheses we were expecting (e.g., the experimental group should have more confidence), we could still see relevant tendencies of novices using the ecological interface showing better understanding and reasoning patterns of the data shown to them in context. Different designs of visualizations with different variables can be further evaluated in the future, incorporating more and various perspectives of interface users in neurocritical care. Different scenarios can be tested while also other expertise-relevant measures can be considered. More recommendations for future research are outlined in the next chapters.

Contributions: This work contributes to the understanding of ongoing challenges and the importance of data representation on daily-used interfaces and requires in-depth dissemination to support novices in developing expertise in neurocritical care. While this research represented CWA models together with the exploration of expertise in the context of neurocritical care, there is not only a scientific contribution shown in this dissertation but also how technology and specific interfaces can be improved to support different end-users on varying expertise levels. The overall intention is to contribute to a better healthcare system in which physicians can provide best practices and dedication toward patients’ health states and share their expertise with newer generations. In the long run, this also has an educational and economic impact, improving hospital resource allocations.