This doesn’t seem like a Tiled question at all xP
You shouldn’t worry about the exact sizing - you can and should scale the art based on the user’s screen size. For a pixel art game especially, you should expect to scale up - pixel art at a native size of 1920x1080p is huge, and will not look nice on any screen other than that size; with a much smaller target resolution, you can scale up to a wider variety of screen sizes. In addition, not everyone uses a 16:9 screen, so you should have a safety margin as you may sometimes need to show a little more on the sides or on the top and bottom.
Even considering room sizes in terms of tiles isn’t necessarily helpful, as the visual scale of tiles can also vary - in some games, a single tile is just a bit of texture, while in some, a single tile is an entire town.
Rather than thinking about room sizes first, think about gameplay and visual density. How much stuff do you want to fit on the screen, and how do you want this stuff to relate to the tiles? Choose your asset sizes, room sizes, and room-to-screen ratio accordingly. Blocking out some rooms with generic placeholder tiles can help you figure this out. In Tiled, you can set tiles to fit the map’s grid size instead of using its native size, so you can use almost any tileset for prototyping.
Other people can’t really help you with this decision, as the appropriate density for your levels depends on your desired gameplay and visual style. There’s no getting around making some rooms (whether as playable prototypes or drawn mockups) and iterating over different possibilities.
One generic piece of advice I can give is that you generally want the player to see the next point of interest from where they are now. A point of interest can be an interactive element, the door to another room, a unique visual, etc. Players will also tend to go to the next point of interest they see, so if you want them to make decisions, you need multiple points of interest to be visible at once.
Escape-the-room games specifically usually work in terms of screenfuls - a single room may consist of multiple connected “screens”, but generally you don’t have to scroll/pan much to see everything on a given “screen”. Even if the game is top-down rather than the more typical first-person view, this still generally applies. The idea is that the player can see most of the points of interest at once, so they can spend their mental energy on working out the connections between the different elements instead of on trying to remember what they were just looking at. Players should generally not have to think about too many connections between screens simultaneously - one or two at a time is fine, more than that will feel overwhelming - but of course, many connections in sequence is perfectly fine and can make your world feel more complex without overwhelming the player as easily.
It’s also common for screens to have sub-screens, so you don’t have to fit all the details on the screen all at once. For example, a radio may appear as just a small radio sprite in the room, but will pop up a larger overlay when the player interacts with it, and this overlay will have all the little fiddly bits the player can interact with.
