The aim of this research is to identify health professional design and uptake recommendations for the GUI of an AI-enabled predictive risk tool. The primary and secondary objectives are described in the following sections.

A total of three 2-month research phases have been scheduled from mid-2023 to mid-2024, with 2 co-design workshops scheduled per phase, one in Adelaide (University of South Australia city west campus), the other in Tonsley (Tonsley Innovation District) for Central Adelaide and Southern Adelaide Local Health Network staff, respectively. If required, participants may attend using teleconferencing software (eg, Zoom) and collaborative software (eg, Miro). Each workshop will be led by an experienced facilitator and run for approximately 2 and a half hours. A total of 2 support staff will also be in attendance: one will observe and note proceedings while a second will provide operational support to the facilitator (eg, information technology troubleshooting, handing out and collecting activity sheets).

Members of the PreHaRM clinical implementation team (CIT) will be asked to promote each workshop by circulating an invitation to colleagues within their peer network. Individuals who confirm their interest in the workshop will be emailed workshop details (ie, session date, time, and location). A total of 6 participants will be enrolled in each workshop, resulting in a total of 12 participants per research phase, for a maximum total of 36 participants (Table 1). The total number of individuals may be less than 36, as participants will be allowed to attend more than one workshop should they choose to; however, this repeat attendance is not a requirement and will be made at their discretion.

Nurses have been identified as the target user of the GUI, and as such, the inclusion and exclusion criteria below reflect the need to attract a diverse range of participants from this professional group; for example, nurses of all levels (including enrolled, registered, and nursing unit managers) will be sought so the diversity of their experience can benefit the design. Criteria for enrollment are listed in Textbox 1 below.

An iterative co-design process will run across each of the workshops. Each workshop includes predesign and generative activities, noting that these activities will culminate in the testing activity held in the third workshop. Figure 1 illustrates the focus and relationship of each workshop, as well as the overall iterative co-design process.

A description of each workshop activity is provided in Table 2 below.

Workshop tools and the method for analyzing their respective results data are described in the follows sections.

Following the completion of each workshop, attendees will be offered the opportunity to participate in voluntary, one-on-one, unstructured interviews with NM. The purpose of the interview is to enable participants to provide further information about their workshop feedback or to provide new feedback if they feel uncomfortable raising it in a group context. Interviews will be offered through teleconferences only (Zoom; version 5.12.8; Zoom Video Communications Inc). These interviews will be offered as an opt-in opportunity, and it will be explained that participation in any workshop is not contingent on their participation in the interviews.

In unstructured interviews, it is unclear what participants may choose to discuss; consequently, inductive thematic analysis will be used to examine participant interview transcripts. This will be performed using NVivo. Participant perspectives will be coded by NM and examined by KCC.

All workshops and follow-up interviews will be audio recorded using a digital recording device; scenario testing activities will be recorded using video cameras; hardcopy artifacts (ie, visualizations and dashboards) will be collected by PreHaRM researchers; and participant surveys will be collated using web-based software. Audio recordings will be transcribed using voice recognition Rev (Rev.com, Inc); video recordings will be reviewed using Adobe Premiere Pro (version 23.0.0; Adobe Inc); Microsoft Forms will host each survey and automatically collate responses. Once the survey period has ended, collated responses will be downloaded. All electronic data will be stored in password-protected folders at the University of South Australia, Australia. All artifacts will be stored in facilities at the University of South Australia, Australia. All data will be retained for a minimum of five years after project completion in accordance with Australian State and Commonwealth guidelines.

Peer-reviewed manuscripts, posters, and oral presentations detailing the co-design process and results will be produced; information used in these research outputs will be deidentified. If video recordings are used, then participant likenesses (audio and visual) will be obscured. Data sets generated and analyzed during this research are not publicly available due to dissemination restrictions resulting from the ethics approval process.

A participant information sheet and consent form will be emailed to all participants before each workshop. Hardcopies will be available at the start of each session, and all individuals must read the information sheet and sign the consent form before participating. This research method has human research ethics approval from the South Australian Department of Health and Wellbeing (#2022/HRE00131) as well as from the Human Research Ethics Committee of the University of South Australia (application ID#204143). No research activity will commence until approvals from both committees have been received. All research data will be deidentified by researchers. Participants will be offered an AU $50 (US $32.82) gift card for their attendance.

The key strength of this research and the broader PreHaRM project is the purposeful engagement of end users in each research activity; this is research with health professionals for health professionals. Preliminary engagement with the state government and local health networks has been achieved through engagement with the CIT. Comprised of senior hospital staff and government health care data professionals (managers and stewards), this group provides strategic information on data collection and use within the local health care system. This insight provides PreHaRM researchers with a deeper understanding of the administrative and technical context in which health professionals work. Similarly, the PreHaRM TAG includes leaders in AI software development and interface design. This group guides researchers’ thinking on the capacity of the PreHaRM algorithm. Both the CIT and the TAG have already been central to the predesign phase [20] and will continue to remain involved in the upcoming stages of the design process. Health professionals drive the generative phase of the PreHaRM tool primarily through their efforts in the “notifiers” and “format” activities, but also through their descriptions of software use and their feelings toward AI within health care. The evaluative activities of researchers and the presentation of these results back to health professionals further involve them in the iterative and generative process at the crux of this research. The broad contextualization of the health care system and guidance of AI development from the CIT and TAG, respectively, are operationalized through the generative activities of health professionals at the “coal face.”

As primary end users, nurses are the target cohort for each workshop, and their input in these sessions will inform the PreHaRM GUI. Within the hospital context, however, it is possible that other health professionals may also use the PreHaRM GUI. Callen et al [21] identified that differences in organizational culture along professional lines represent differences in attitudes toward HITs. Therefore, the focused recruitment of one health professional group is a potential limitation; future research should broaden recruitment to capture the design requirements of additional health professional cultures. Additional in situ research should be conducted with health professionals to confirm the issues they noted in the sessions are as consequential as they are real and to seek further refinements that improve the adoption of the PreHaRM tool.

This project will engage health professionals in an iterative co-design process across 3 workshops. This process of repeat engagement will enable researchers to refine recommendations for the aesthetics and functionalities of the PreHaRM GUI. It is anticipated that by adopting their recommendations, health professionals will be more likely to accept and use this tool and to prevent forecast events from occurring. The target audience of this protocol is other health researchers, health professionals interested in research, and user interface and user experience design researchers.

The PreHaRM tool will combine an AI-enabled predictive risk algorithm with a co-designed GUI. The purpose of this algorithm is to identify the potential occurrence of adverse events, whereas the purpose of the GUI is to convey these potential events to health professionals in a meaningful manner. Indeed, the presentation of the algorithm results is paramount if the warning it displays is to be acknowledged and actioned. Co-design will be used in this research to identify end user aesthetic and functionality design recommendations to help ensure the appropriateness of the final product and its incorporation with existing work practices. This research takes as a basic assumption that the PreHaRM GUI, designed with input from health professionals, will be used by professionals to the benefit of patients and their families.