We have one emeritus faculty member who was using an Apple IIe with programs he wrote for it many, many years ago to analyze data up until just a handful of years ago. I actually put a new PSU in it at one point. I think he would have had a legitimate case for using a IIe emulator if he'd wanted, especially since he wasn't interfacing with hardware. He finally gave up on that research project(the computer is in my office now).
I’m just curious, if I were to set up a brand new, modern lab doing the sort of applications you describe, would the manufacturers set me up with equipment that requires terribly outdated computers and software to operate them? It would strike me as odd that development of scientific and medical instrumentation had somehow stopped cold in the 1990s...
I would say that the current versions of stuff I deal with connect either via LAN or USB. That probably covers 95% of major scientific equipment, with the fringe cases using proprietary interfaces.
Often times, older interfaces stick around past their prime just because of design inertia and a need for continued compatibility with even older stuff(remember that in an interconnected system, you may only replace part of it and retain older, still functioning parts).
As an example, for a good many uses, the interface I call HP-IB(Hewlett Packard Interface Bus), but that is also called GP-IB(general purpose interface bus) or IEEE-488 mostly went out of fashion in the 1980s. I have a Varian GC-MS that was bought new around 2003 that uses it, along with a Beckmann Capillary Electrophoresis instrument bought in ~2005 or so that uses it via an IEEE-488 bridge to USB bridge(not an option for something like a GC-MS, not the least of which because the software support isn't there, but also technical considerations). Agilent(formerly Hewlett Packard Instruments) started transitioning to LAN in the late 90s/early 2000s. I know Finnigan instruments-one of their big competitors(now part of Thermo-Fisher)-started going to LAN in the early 90s, but even around ~2000 they were still using thin net 10base2 and the actual communication protocols can twist your head into knots(I had to spend a while hunting on Ebay for a particular model and revision of 3Com ethernet card that's needed to get one mass spec I use to work). There was also plenty on RS-232, even up to maybe ~10 years ago.
I wish I could locate the paper again, as I seem to have lost it, but a while back I came across a nice summary of different instrument interface buses that compared speed, latency, and a few other factors. It essentially concluded that where latency isn't an issue-as is the case with some types of simpler data analysis where the instrument itself can store everything and then offload it as needed-USB is a good choice(essentially anywhere RS-232 would have been historically used) while LAN is better when you basically need real-time minimal latency data collection directly to the computer. Depending on the exact set-up, this may also be where proprietary interfaces come into play(the OS 9 NMR I mentioned uses that, while Varian/Agilent systems use LAN in RHEL and I've worked with others with their own proprietary interface in various OSs).
It's also worth noting that HP-IB is, in a lot of ways, almost an ideal interface for certain applications(MS use is one of those) since it's fast enough at the rate for a typical data collection rate with even less latency than LAN. The cost to implement and the physical size of the connectors are its two biggest drawbacks.
And, yes, as things get even more complicated, proprietary is still often the way to go. We have a 20 year old MALDI-TOF-MS that is loaded full of proprietary interface cards in the computer. That computer needs to control a lot of things, then collect a whole lot of data fast. I actually am not sure what more modern systems use, but I would not be surprised if it's still proprietary. I was eying a mothballed GCxGC-TOF-MS not too long ago, and one of the things that scared me off aggressively pursuing it was that no one could remember if the computer/specialized interface cards worked in that particular one.