While the previously mentioned artifacts can be considered to be the status quo, they will likely be replaced or complemented in the future by pervasive computer-based solutions. Advances in the development of rugged mobile devices, wireless internet access and ad-hoc networking as well as a changed perception of the overall safety situation in the western world led to increased research and development activities. Prototypes have already been tested in the field and particularly proven to work with respect to technical or infrastructural requirements like reliability of devices or data transmission in the field (e.g. [22]). Some of them process and visualize a wide range of information, e.g. patients’ conditions, hospital assignments, or chronology (Fig. 5).
Questions of usability in this domain are rarely related to cognitive ergonomics, task analysis, human-computer interaction, or design (e.g. [22, 23]). Following some more general remarks to usable computer-based solutions in MCI management, challenges and approaches to cognitive ergonomic design of interactive tables, maps and note-taking are considered.
General remarks on usability
Developing interactive cognitive artifacts for the context of pre-hospital medical care is a challenge for various reasons. Interface and interaction should be designed iteratively and by user participation. However, short-term scheduling is complicated by work schedules and staff requirements.
Field studies and test runs can hardly be projected. Workshops, interviews and expert reviews often take place in conference facilities or office rooms. Such favorable conditions differ completely from the real context of use. Thus, natural environments, interruptions and other performance disturbing factors that characterize MCIs have to be considered specifically and carefully. What seems like an appropriate approach in a relaxed training situation might be a hardly manageable case of information overload in the field. Furthermore, computer-based tools and systems are an additional medium of communication and channel of information. They have to be aligned with established workflows and organizational structures which depend on thorough exchange of radio messages.
Practice in efficient and safe handling as well as operating interactive cognitive artifacts cannot be ensured by exercises or training courses alone. It can only be derived from intense and regular application. Because MCIs are rare events for single EMS employees, this requirement can only be met, if interactive systems will be used in daily routine and not just during MCIs. A rugged tablet PC (Fig. 6) can be considered an appropriate hardware solution for a consistent user interface designed for regular and extraordinary missions. We have basically confirmed this in formative and summative evaluations based on a prototypical application [4, 5].
Because of possible difficulties arising from wearing gloves, touching accidentally or working with dirty hands, pen-based interaction should be supported by the system. Losing the pen might be an additional risk but it can be minimized by fastening it with a tear-resistant but flexible ribbon. Other input methods which are supported by state-of-the-art and off-the-shelf tablet PCs, e.g. speech or gestures, have to be judged critically. One the one hand, speech input usually demands high working memory resources of users and recognition rates could drastically decrease in noisy environments like MCI settings. On the other hand, gestures have to be remembered and performed correctly. This might be challenging with respect to weather or physical conditions.
Among other aspects, consistency can be ensured or improved by using well-known and appropriate layouts, data input widgets, visualizations, feedback mechanisms, error messages, symbols and colors. Although striving for it, certain screen layouts and interaction elements, which are important to MCI management, will not be used by incident commanders in their daily duty as regular emergency physicians or paramedics. One example is the distribution of triage categories shown in Fig. 6. Especially while dealing with well-known design issues for these parts, well-established design principles (e.g. [24]) should be considered. These best practices need to be applied to the specific context of MCIs [25]. As they represent essential entities and relationships, currently used cognitive artifacts might be an appropriate starting point for this process.
In safety-and time-critical domains with highly skilled personnel, it seems advisable to build upon practical experiences. Furthermore, questions of automation, adaption, and individualization arise with respect to division of tasks between human and machine. User interfaces and interaction methods have to be designed with respect to hardware capabilities (e.g. screen size, resolution, input and output modalities) and context.
Challenges and approaches to ergonomic design
As mentioned before, incident commanders currently take private notes and work with various paper-based documents in order to enhance their SA. Some of them are stationary and others have to be delivered manually. They can be arranged, sorted, or marked by the user to a limited extent. Computer-based tools allow data access and exchange near real-time and from remote but offer only limited screen space. Feedback about incoming data must be given explicitly.
Forms and tables
With respect to forms and tables, some major challenges are
-
efficient navigation between different sections;
-
comprehensive visualization of larger datasets;
-
fast browsing of numerous datasets.
Tabs, i.e. multiple screen masks within a single container and a navigational widget, can be used to group single tables logically and ease access. Another approach is to mark an item in one table and see links to related items in other forms. This requires efficient design solutions for backward and forward navigation, e.g. breadcrumbs.
Larger datasets might not be presentable in a single table row. Possible approaches would be folding out entries on demand or displaying detailed data of a marked entry in an overlay panel. Search options have to be available and should support phonetic search. By doing this, search results would be returned that sound similar to the given term. This might be important because incident commanders might not know correct spelling of some search terms, e.g. last names, or make typing errors while having serious time pressure. Moreover, in favor of searching by entering data freely, filter mechanisms should be implemented, if the range of values is limited. They would only offer proper values. Activated filters must be clearly visible, e.g. by changed background colors or other visual hints. Otherwise, the subset of displayed datasets could be perceived as the total set. Temporarily marking favorites, e.g. in terms of patients to remember, can be a feature to relieve incident commanders’ working memory and allow them to continue tasks later on more easily.
With respect to browsing larger datasets, paging and continuous scrolling are basic options. While the latter one is the most prevalent at mobile devices in general, we observed some difficulties with pen-enabled devices. Some users slipped of the surface while trying to move the pen up or down. Such difficulties could increase stress levels. Therefore, we recommend implementing a more fail-safe paging solution–at least in addition to the other one. Regardless of the approach, all interaction elements should offer a sufficient target area in order to deal with pen interaction challenges like occlusion.
Charts and maps
The number of patients in specific triage category is one of the most important chunks of information for incident commanders. In addition to tabular or textual visualizations, bar charts can be an appropriate design solution. During our workshops we presented four drafts differing in the number of categories and labeling (Fig. 7) to 36 EMS employees. We asked for their favorite or a self-created version. While 2 responses were ambiguous, 34 could be evaluated. The 3 most preferred versions got 9, 6 and 5 votes, respectively. 5 participants created their own solution. Allowing for minor modifications, e.g. position of labels, one version got 14 votes. The drafts were subject of controversial discussion. They enabled us to better match our conceptual model with their mental model beyond the use cases associated with the actual chart.
Using location-based services by tracking casualties, EMS employees or vehicles and visualizing them on maps looks promising but raises issues of reliability. Both tracking inaccuracies (e.g. in buildings, under bridges) and loss or removal of locatable items (e.g. by casualties in a state of shock) could result in inaccurate data. Such data would be worse than none at all. More or less stationary information (e.g. territorial allocation of operation areas) can be represented more reliable (Fig. 8).
Data exchange with geographic information systems in emergency control rooms or command vehicles might be necessary. Moreover, some incident commanders and EMS managers expressed reservations about modifying situation maps by multiple actors. They were strongly in favor of a read-only mode for co-workers in the field.
Personal notes
Personal note-taking should always be possible and could be realized easily on a pen-enabled tablet PC. A digital notepad within reach of every screen mask can be a first step. It should support both handwriting recognition and freehand drawing. Advanced solutions could offer more shortcuts, e.g. for creating tables or marking entries in different colors, or support annotations of pre-defined user interface components, e.g. a digital triage tag with notes about the patient.