Open-source software may unify the medical-records realm (ComputerWorld) [LWN.net]

Steven J. Vaughan-Nichols takes a look at open source medical software. "One EHR [electronic health records] system, however, does have a proven record, since its introduction in 1982: VistA (Veterans Health Information Systems and Technology Architecture), the U.S. Veterans Administration's public-domain EHR. VistA has become the foundation for over a dozen proprietary and open-source medical record software suites."

to post comments

Open-source software may unify the medical-records realm (ComputerWorld)

Posted Sep 15, 2009 21:17 UTC (Tue) by nix (subscriber, #2304) [Link] (26 responses)

No mention at all of the eye-burning qualities of VistA's code, I see.

I downloaded it after the last LWN article on it, and, y'know, the
snippets of code that were posted in comments then were not
unrepresentative at all. It really *is* all that bad.

Open-source software may unify the medical-records realm (ComputerWorld)

Posted Sep 15, 2009 21:32 UTC (Tue) by Cardinal_Bill (subscriber, #23688) [Link] (24 responses)

Low bid contract, unless it was done sole source in which case it was a high no bid contract. Your tax dollars at work.

Open-source software may unify the medical-records realm (ComputerWorld)

Posted Sep 15, 2009 21:57 UTC (Tue) by horen (guest, #2514) [Link]

There's quite a lot of interesting history to the VISTA package. I suggest you read the VistA History page, and also the Hardhats website, for some much-needed background.

Bottom-line: Most of the work was done after-hours, by people with a solid agenda for positive change and advancement.

Open-source software may unify the medical-records realm (ComputerWorld)

Posted Sep 15, 2009 23:19 UTC (Tue) by jordanb (guest, #45668) [Link] (11 responses)

VistA was developed by employees of the VA, mostly as a skunkworks project, because they were committed to trying to make positive changes in the way the hospital system operated -- using computers -- to be more efficient and take better care of their patients. They engaged in action that their counterparts in the private hospital system have thus far been unwilling or unable to take.

I know that the idea that such people could exist in a 'government bureaucracy' is so fundamentally against the Libertarian/Ayn Rand/public-choice-theory religion's terminally cynical view of civil servants, so I have no doubt that many here will try to denigrate the whole effort so as to avoid cognitive dissonance with their worldview.

VistA was written in MUMPS in the 1970s. MUMPS is designed to be an incredibly brief language and given computer resources of the time, brevity was considered a virtue of 'good code.' People in some sectors of computing still today fetishize low character count in source code.

That said, the code of VistA does seem to be very difficult to read, and therefore maintain. I said in that last discussion, though, that most of the single-letter items (according to wikip) are language keywords that it might be possible to mechanically expand.

Open-source software may unify the medical-records realm (ComputerWorld)

Posted Sep 16, 2009 0:32 UTC (Wed) by Doogie (guest, #59626) [Link] (1 responses)

Business Week had an interesting article about the VA's use of technology to substantially improve its level of care:

According to a Rand Corp. study, the VA system provides two-thirds of the care recommended by such standards bodies as the Agency for Healthcare Research & Quality. Far from perfect, granted -- but the nation's private-sector hospitals provide only 50%. And while studies show that 3% to 8% of the nation's prescriptions are filled erroneously, the VA's prescription accuracy rate is greater than 99.997%, a level most hospitals only dream about. That's largely because the VA has by far the most advanced computerized medical-records system in the U.S.

Our tax dollars (and open source) at work, indeed. I don't see why people could think this kind of an improvement is a bad thing.

Open-source software may unify the medical-records realm (ComputerWorld)

Posted Sep 16, 2009 13:21 UTC (Wed) by drag (guest, #31333) [Link]

> Our tax dollars (and open source) at work, indeed. I don't see why people could think this kind of an improvement is a bad thing.

Well if its a skunkworks project then that mostly means it's going to be something people do mostly on their own time. So "yay" open source, the government did not have much to do with it, at least originally.

Striking a nerve?

Posted Sep 16, 2009 16:18 UTC (Wed) by clugstj (subscriber, #4020) [Link] (8 responses)

"cynical view of civil servants"

WARNING: Sweeping generalities ahead:

Dude, have you ever gone to the DMV? Government bureaucracy is fundamentally broken because it exists mainly to perpetuate itself, not to serve any worthwhile purpose. From what I've heard, the VA is not special in that respect. Of course there are good people there and I'm happy that they've been able to do some good, but overall, the price paid for the service rendered is always higher from the government than the private sector.

Striking a nerve?

Posted Sep 16, 2009 18:02 UTC (Wed) by justme (guest, #19967) [Link]

It seems that the numbers above disagree with you. No one passed a law preventing insurance companies from obtaining a 99.997% prescription accuracy rate, or private software companies from developing adequate medical records software. But it hasn't happened in the private sector, only at the VA. Why? Maybe because what the private sector will always do better is generate profit. Good for them, but that's not what I go to the doctor for.

Striking a nerve?

Posted Sep 16, 2009 18:22 UTC (Wed) by flewellyn (subscriber, #5047) [Link] (1 responses)

Yeah, except none of those things are actually, y'know, true.

Striking a nerve?

Posted Sep 17, 2009 1:29 UTC (Thu) by jimmybgood (guest, #26142) [Link]

Well, you beat me to it. I've used the VA. I simply don't believe the prescription accuracy rate is 99.997%, so discussing why it is is a waste of time. It isn't.

Striking a nerve?

Posted Sep 16, 2009 23:04 UTC (Wed) by foom (subscriber, #14868) [Link] (2 responses)

> Dude, have you ever gone to the DMV?

Sometimes I feel like I must be the only person in the country who doesn't have nightmares about going to the DMV...Whenever someone talks about how bad government is, they *always* bring up the DMV.

Heck, give me the DMV over FedEx any day! I just love it when the drivers do a tag-and-run, without so much as knocking. One time when I called and complained (about 10 seconds after watching the truck drive away out my front window and finding a sticker on my door), the operator told me "That's not possible, why would the driver stick a tag on the door without knocking, they'll just have to come back tomorrow and try again, after all! It doesn't make any sense!" Which would be a very nice argument, indeed, if they had made it to the driver instead of me.

Striking a nerve?

Posted Sep 17, 2009 3:54 UTC (Thu) by jordanb (guest, #45668) [Link] (1 responses)

FWIW, last time I got my license renewed I went to the DMV at the Thompson Center in Downtown Chicago. They were efficient and friendly. The whole process took maybe an hour and the lady even made pleasant smalltalk with me as she worked the computer.

Striking a nerve?

Posted Sep 17, 2009 5:59 UTC (Thu) by k8to (guest, #15413) [Link]

Last time i went in person to get my license renewed it was pretty awful... but the two times before that, it was a 15 minute process. Ding Ding Ding. Thanks, bye.

Striking a nerve?

Posted Sep 17, 2009 15:10 UTC (Thu) by nye (guest, #51576) [Link] (1 responses)

>overall, the price paid for the service rendered is always higher from the government than the private sector.

That must be why the US health-care system is renowned the world over for being cheap and effective. Thanks for clearing that up.

Striking a nerve?

Posted Sep 17, 2009 16:42 UTC (Thu) by dlang (guest, #313) [Link]

it depends what you are looking for.

effective, it's very effective at finding new solutions (these are frequently very expensive)

cheap, parts of it, but nto the parts that overlap with the new solutions.

Open-source software may unify the medical-records realm (ComputerWorld)

Posted Sep 16, 2009 0:04 UTC (Wed) by nix (subscriber, #2304) [Link] (10 responses)

What? No, apparently MUMPS code *always* looks like this... there's a
culture of "if you can abbreviate, you should abbreviate".

Open-source software may unify the medical-records realm (ComputerWorld)

Posted Sep 16, 2009 1:42 UTC (Wed) by proski (guest, #104) [Link] (1 responses)

I'm scared to think how MEASLES code would look like

Open-source software may unify the medical-records realm (ComputerWorld)

Posted Sep 16, 2009 2:39 UTC (Wed) by flewellyn (subscriber, #5047) [Link]

It's the pattern of dots you find on the paper tape used by a Turing machine.

Open-source software may unify the medical-records realm (ComputerWorld)

Posted Sep 16, 2009 4:56 UTC (Wed) by eru (subscriber, #2753) [Link] (7 responses)

Ever seen APL programs? Now that is the gold standard for brevity. No keywords, just special operator characters, and almost everything is done with matrix and vector values flowing though expressions, so a single expression often does what a 100 line C function would do. To not spoil this style, programmers of course tend to use short variable names on APL...

The amazing thing is APL can be efficient as well. Back in 1985 or so I worked at a tiny company that made graphics software for running on a PC (XT and later AT). Most of the code was in interpreted APL (which I didn't touch, my job was just accelerating the rendering by assembly-language routines). APL allowed the code to be very dense, leaving more memory for the data, and its matrix operations suit graphics very well.

Open-source software may unify the medical-records realm (ComputerWorld)

Posted Sep 16, 2009 9:43 UTC (Wed) by nix (subscriber, #2304) [Link] (6 responses)

Well, yeah, that's what happens if you have a language used by mathematicians!

(I was recently privileged to see some C code written by the designer of APL. He writes C the same way: single-letter variable names, minimal comments, a pile of macros that transform any longer function calls into single-letter operations, and all of them heavily nested.)

Open-source software may unify the medical-records realm (ComputerWorld)

Posted Sep 16, 2009 10:56 UTC (Wed) by nye (guest, #51576) [Link] (3 responses)

But *why*? It's not like readable variable names have a runtime penalty. I don't get why some people associate 'short source code' with 'short object code'.

Open-source software may unify the medical-records realm (ComputerWorld)

Posted Sep 16, 2009 12:16 UTC (Wed) by eru (subscriber, #2753) [Link]

I don't get why some people associate 'short source code' with 'short object code'.

In some cases it might be old habits formed with old language implementations... In the APL version I mentioned earlier, short source code did in fact have a better performance than longer one, because it was a direct interpreter (a long identifier needs more bytes to store it, and a longer string comparison to match it). The same was also true of most old BASIC implementations.

Open-source software may unify the medical-records realm (ComputerWorld)

Posted Sep 16, 2009 12:29 UTC (Wed) by ajb (subscriber, #9694) [Link]

Well, there is a reason mathematics uses single letter variables, but I'm not convinced it applies to code. In maths, it's often useful to be able to see spatial relationships between the occurrences of variables in an equation, and this is harder for multi-character names.

Open-source software may unify the medical-records realm (ComputerWorld)

Posted Sep 16, 2009 14:35 UTC (Wed) by ballombe (subscriber, #9523) [Link]

Interpreters written by mathematicians used to have the property that long variable names incurred a run-time penalty, so it is not surprising they expected as much.

C Code by the APL designer

Posted Sep 17, 2009 23:51 UTC (Thu) by pr1268 (guest, #24648) [Link] (1 responses)

Sounds like a contestant in the IOCCC. Is he? ;-)

C Code by the APL designer

Posted Sep 20, 2009 19:35 UTC (Sun) by nix (subscriber, #2304) [Link]

Amazingly, no. But it should have been.

And this compares with closed source ... how?

Posted Sep 15, 2009 22:59 UTC (Tue) by felixfix (subscriber, #242) [Link]

Just because the available open source software looks lousy doesn't mean the closed stuff is beautiful. I have seen tons of closed proprietary source and none of it is beautiful. I doubt the open source stuff is the worst.