Usability testing with a prototype user interface of an Artificial Intelligence driven air-Safety Tool (AISaT)
This paper reports on a usability study of a prototype AI-driven air-safety tool (AISaT) involving ten hospital staff members, which identified specific interface improvements and highlighted critical decisions regarding data variables, infection assumptions, and user autonomy needed for future development.
Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of a preprint that has not been peer-reviewed. It is not medical advice. Do not make health decisions based on this content. Read full disclaimer
Imagine you have a super-smart digital assistant designed to solve a very specific problem: keeping hospital rooms safe from invisible germs floating in the air. This tool is called AISaT (Artificial Intelligence driven air-Safety Tool). It uses complex math and computer science to figure out exactly where to place a giant air filter (like a HEPA filter) so it catches the most germs before they can make anyone sick.
But before the doctors and nurses could use this "smart assistant," the creators wanted to make sure it wasn't too confusing to use. They invited 10 hospital staff members (doctors, nurses, and building managers) to look at a prototype (a practice version) of the tool and tell them what they thought.
Here is what happened, explained simply:
The "Test Drive"
The researchers didn't let the staff use the tool themselves yet. Instead, they showed them the screen on a video call and said, "Watch this, and tell us what you're thinking out loud." They walked through six steps, like filling out a digital form about a hospital room:
- The Room: How big is it?
- The People: Who is in there? (Patients, doctors, etc.)
- The Filter: Where should we put the air cleaner?
- The Result: A 3D picture showing how well the filter works.
What the Staff Said (The Good, The Bad, and The Confusing)
The staff gave a lot of helpful feedback, which the researchers grouped into a few main categories:
1. "What am I supposed to do?" (Understandability)
- The Problem: The instructions were a bit like a map written in a foreign language. Staff weren't sure if they were supposed to put five filters in the room or just pick the best one out of five suggestions.
- The Fix: They suggested renaming confusing terms. Instead of "Human Models," call them "Patient" and "Doctor." Instead of "Mitigation," just say "Air Filter." They also wanted clearer labels so they knew they were comparing two different spots.
2. "Is this real life?" (Content & Reality)
- The Problem: The tool was a bit too simple. It didn't know about things that actually exist in a room, like windows, curtains, or furniture blocking the air. It also struggled with the question: "Who is sick?"
- Some staff worried that if the tool assumes everyone is sick, it would recommend buying thousands of expensive filters that the hospital can't afford.
- Others worried that if it assumes no one is sick, it might miss a dangerous patient.
- The Fix: The tool needs to be smarter. It should ask about specific diseases (like Flu vs. TB) and account for things like coughing or heavy breathing, not just "breathing level."
3. "I can't find the buttons!" (Navigation)
- The Problem: Moving the little digital people around the room was hard. You had to press keyboard keys like "Q" to move them, which felt clunky. It was like trying to drive a car using only the horn.
- The Fix: Let users drag and drop items with a mouse, just like moving furniture in a video game. Also, some people felt overwhelmed seeing all the steps on one giant page; they preferred seeing one step at a time.
4. "I can't see the difference!" (Visibility)
- The Problem: The tool used colors (like red and blue) to show where the germs were. But for people who are colorblind, or for screens where the colors clash, it was impossible to tell the difference between "safe air" and "dangerous air."
- The Fix: Use shapes, numbers, or clearer pictures instead of just relying on color. Also, make the text bigger.
5. "Will this actually fit in my room?" (Workflow & Practicality)
- The Problem: The tool might say, "Put the filter right here!" But in real life, that spot might be right next to a door, or there might be no power outlet nearby, or the cord would be a tripping hazard.
- The Fix: The tool needs to check if a spot is practical, not just scientifically perfect. It should ask, "Is there a plug nearby?"
6. "Who is the boss of this tool?" (Ownership)
- The Problem: Nobody was sure who should be in charge of using it.
- The Clinicians (Doctors/Nurses) said: "We are too busy seeing patients to measure rooms and input data."
- The Estates Team (Building Managers) said: "We know the room sizes, but we don't know which patients are sick."
- The Conclusion: It might need a team effort. Maybe the building manager sets up the room size once, and the nurse just updates it when a new patient arrives.
The Big Takeaway
The paper concludes that while the AISaT tool has a brilliant idea, it needs a "software update" to become user-friendly.
The researchers learned that:
- The instructions need to be simpler.
- The visuals need to be clearer (less confusing colors, more realistic pictures).
- The data needs to be more realistic (accounting for real furniture and real-world constraints like power outlets).
- The process needs to fit into the busy lives of hospital staff, not add more work to their day.
In short, the tool is like a new recipe for a cake. The ingredients (AI and math) are great, but the instructions (the user interface) need to be rewritten so that the bakers (hospital staff) can actually bake the cake without burning the kitchen down.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.