Emergency departments in Germany face significant challenges in managing increasing workloads and improving patient care. Introducing robotic assistance could offer promising solutions for patient admission processes. However, to realize the potential of robotic assistance, we need to design systems that focus on patient needs.

Building on research in human-robot interaction that emphasizes a patient-centered approach, we designed a co-creation approach—a four-phase workshop—to understand patient expectations of robotic assistance. This co-creation approachis part of the RAER project, which aims to improve interactions between robotic systems and patients in emergency department waiting areas.

In this post, we share our preliminary insights from three workshops conducted with patients in August 2024, highlighting the method's suitability for exploring design concepts for robotic assistance.

Conceptualizing a Co-Creation Workshop to Explore the Design Space of Robotic Assistance in the Emergency Department

Our endeavor started with a lot of literature work, and the co-creation approach was strongly inspired by Axelsson et al. (2022). In April 2024, we were already able to do a pilot study with experts in human-robot interaction and interaction design to test and refine our co-creation approach.

In August 2024, we conducted three workshops with 14 participants between the ages of 20 and 60 in a dedicated area within the emergency department of Charité - Universitätsmedizin Berlin.

Conducting the Co-Creation Workshop

As shown in the figure below, each workshop followed a predefined process (with a duration of three hours). Divided into four phases, participants reflected on their experiences, evaluated their experiences from different perspectives, contextualized their expectations for a robot assistant, and placed these expectations within real-world interaction design concepts for robot assistance in the emergency department.

Phase 1: Reflecting on Experiences
In the first phase, we encouraged participants to reflect on their experiences in emergency department waiting areas. Each participant wrote their personal experiences on sticky notes, which they then presented to the group. This activity fostered engagement with the workshop context and set the stage for deeper discussions in the subsequent phases.

Phase 2: Assessing Experiences
Building on the reflections from the first phase, we prompted participants to identify the needs and concerns of actors, including physicians, clinical staff, patients, and other key actors related to emergency department processes. Participants captured their insights on sticky notes and shared them individually with the group to stimulate discussions.

Phase 3: Contextualizing Expectations
In the third phase, we tasked participants with contextualizing their expectations into concrete requirements for robotic assistance in the emergency department. At this stage, we explicitly introduced the workshop’s primary context—the deployment of robotic assistance in emergency departments. Participants were encouraged to contextualize their insights from the second phase by relating them directly to the potential use of robotic systems.

To support participants in their ideation process, we provided three focus areas for consideration:
1. Environment: How should the emergency department be designed to accommodate robotic assistance, and where should such systems be physically located?
2. Appearance: What should the visual design and form of robotic assistance look like to ensure usability and acceptance by patients?
3. Capabilities: What functions should a robotic assistance perform to interact with patients and address their needs individually?

Participants documented their ideas on sticky notes and shared them with the group. To enrich their reflections, we posed targeted questions and provided suggestions on specific aspects, sparking in-depth group discussions.

Phase 4: Situating Expectations in Practice
For the final phase, we divided participants into subgroups of two to three individuals. Each subgroup was given a canvas to sketch interactions between patients and a robotic assistant in an emergency department. We, furthermore, asked them to describe interactions step-by-step, focusing on actions the patient and the robotic assistance might take during a specified interaction, such as admitting patients for treatment or measuring patients’ vital signs.

The subgroups presented their results, discussing whether their expectations from earlier phases had evolved and if their scenarios aligned with the envisioned robotic interactions. This phase concluded with a feedback session to gather participants’ reflections on the workshop’s structure and activities.

Reflections and Preliminary Insights

The co-creation workshops yielded valuable insights into the expectations and needs of patients regarding robotic assistance in emergency department waiting areas. These insights were grouped into five key themes, highlighting the importance of designing robotic systems that address practical and emotional patient care.

Fundamental Patient Needs
Participants consistently emphasized their emotional and physical requirements in the emergency department. They expressed a strong desire for rapid medical attention and relief from pain, alongside a need for reassurance and empathy from staff or systems. Uncertainty about treatment processes and interactions with others (e.g., clinical staff), including safety concerns, further underlined the importance of creating a secure and supportive environment.

Improving Communication
Clear communication emerged as a critical factor in enhancing the patient experience. Participants desired transparency about emergency department processes, such as the next steps in treatment. Additionally, participants suggested using waiting times more effectively to educate patients, for example, through accessible information about their conditions, treatments, or available resources.

Enhancing the Physical Environment
The workshops revealed several ways to improve the spatial design of waiting areas. Participants proposed better wayfinding systems, such as clear signage and intuitive layouts, to reduce confusion. They also advocated for noise reduction to create a calmer atmosphere, particularly during long waits.

Integrating Patient-Centered Technology
Participants expressed clear preferences for how technology should enhance their care. They emphasized the need for barrier-free, inclusive, and multilingual systems, incorporating visual aids to help bridge communication gaps. Furthermore, participants envisioned technology as a supportive companion, offering real-time updates about their treatment status.

Expectations for Robotic Assistance
Participants outlined several key requirements specific to robotic systems. They suggested separating tasks across different robotic assistances, such as terminals, to avoid congestion. For example, one system could provide general information, another could handle patient registration, and a third could give updates about wait times and treatment progress. Privacy and security were paramount, with patients advocating for features like screen privacy filters or personal QR codes for accessing information discreetly. Robustness and durability were further highlighted as critical in the demanding emergency department environment.

Regarding capabilities, participants envisioned robots taking on roles such as initial triage, symptom clarification, and vital sign measurement. However, they stressed the importance of human oversight to ensure flexibility and trust. Participants also preferred robots that assist with administrative and technical tasks, leaving more complex interactions, such as discussing treatment options, to physicians.

Next Steps

The preliminary findings of the co-creation workshops play a central role in the ILLUMINATION project, which explores the use of large language models (LLMs) in healthcare. Inspired by advances in artificial intelligence, ILLUMINATION envisions an emergency room workflow where patients share their symptoms with an LLM-based application instead of filling out traditional forms. This system would assess the symptoms and suggest pre-prioritization to medical staff, streamlining the triage process.

Overall, the workshop results provide critical insights into patients' mental models, preferences, and expectations for novel robotic systems in emergency departments. These findings will guide the design of future systems that promote patient-centered interactions with LLM-based applications.


References
Minja Axelsson, Raquel Oliveira, Mattia Racca, and Ville Kyrki. 2021. Social Robot Co-Design Canvases: A Participatory Design Framework. J. Hum.-Robot Interact. 11, 1, Article 3 (March 2022), 39 pages. https://doi.org/10.1145/3472225