Publication: Investigating User Requirements: A Participant Observation Study to Define the Information Needs at a Hospital Reception
Authors: D. Sommer, T. Greiler, S. Fischer, S. Wilhelm, L.-M. Hanninger, F. Wahl
Published in: International Conference on Human-Computer Interaction (HCII), Springer CCIS 1833, pp. 157–166 (2023)
DOI: 10.1007/978-3-031-35992-7_23
In brief
For many people the reception is their first impression of a hospital – and it shapes the perceived quality of the entire stay. Most people who walk through the door are under some strain: a patient before a procedure, a relative who is worried, a visitor in an unfamiliar and often confusing environment. It is precisely in that moment that it is decided whether someone feels looked after or lost. Yet surprisingly little is known about what information is actually requested there. In our literature search, we could not locate a single study that systematically captured information needs at a hospital reception. This work closes exactly that gap.
In December 2022, we observed the entrance hall of a rural Bavarian hospital for one week and recorded 1,499 requests using a standardised form. The goal: understand who comes to the reception with which questions – as a basis for designing service robots.
Why the reception is an underrated place
The reception is more than a desk in the entrance area. It is a hub, a signpost, the first point of contact, and not infrequently an emotional buffer as well. This is where paths converge, where uncertainties are absorbed, where it is decided whether a person takes the right lift or spends a quarter of an hour searching the wrong wing. This work is demanding because it constantly shifts between routine and exception – and at the same time it is hard to grasp, because it usually goes undocumented. No one writes down how many times a day the same directions are requested.
That is exactly why the sober look is worthwhile: before talking about technology, one has to understand the work that actually takes place. A requirements analysis does not begin with the question of what a robot could do, but with the question of what people really need in this place. Participant observation – the patient listening and counting over a full week – provides a more reliable foundation for that than any assumption made from a distance.
Who asks what?
Requests break down clearly by group:
- Visitors: 51.3 % (n = 769)
- Patients: 38.5 % (n = 577)
- Employees: 5.3 % (n = 79)
- Other: 1.6 % (n = 24)
For visitors, the most frequent matters were showing a COVID-19 test certificate (n = 289) and asking for a patient's room number (n = 204). Patients mostly asked about appointment registration (n = 148), the procedure in an emergency (n = 98) and orientation within the building (n = 79). Employees had more administrative requests such as borrowing keys (n = 39). On average, staff needed 65 seconds to attend to one person.
What is striking is not only the frequency of individual questions, but their distribution. More than half of all requests come from visitors – that is, from people who do not know the building and, as a rule, do not enter it regularly. On top of that, most questions cluster around a manageable number of recurring topics: directions, room numbers, registrations, procedures. The short average handling time underscores this finding. Much of what is done at the reception is quickly answered – but it adds up over the course of the day to a considerable, continuous load.
Why it matters
The finding is unspectacular and precisely for that reason valuable: a large share of reception work consists of a few clearly defined, repetitive standard questions – directions, room numbers, procedures. That is exactly the task profile a service robot can handle reliably, freeing staff for more complex and personal concerns.
The direction of that relief matters. The point is not to replace the person at the reception, but to free them from mechanical repetition. Someone explaining the same route for the twentieth time has less attention left for the anxious relative who actually needs a conversation. If a system takes over the predictable standard enquiries, staff gain time for exactly those situations in which human attention cannot be substituted. A clean division of labour begins with knowing the tasks in the first place – and that is precisely what this study provides.
The basis for user-centred robotics
This study is the starting point of the series: first we understand the real need, then we design the solution. The requirements gathered here fed directly into the reception and orientation functions that we later evaluated in live operation. At Athegus this principle applies across the board: robotics guided by actual user needs – not by what happens to be technically feasible.
This ambition has consequences for development. A solution that grows out of observing real requests meets the need more precisely, is more readily accepted, and delivers value from day one – because it answers exactly the questions that are being asked anyway. The requirements analysis described here thus becomes not an abstract specification document, but a clear, verifiable starting point for design. For us, this is sovereignty in practice: technology that follows people, not the other way around.