Monday, February 23, 2009

Computerizing Medical Records: What Could Go Wrong?

The other day I was in my dentist's office getting my teeth cleaned (wait, it gets better). Like most other health professionals, my dentist keeps patient records on paper in file folders bearing multicolored tabs, all crammed into shelves behind the receptionist's desk for easy access. As I discussed the bill with the office's insurance person, I glanced behind her and saw the name of a friend scrawled along a protruding edge of one folder.

Now, technically, that was a breach of confidential medical information, I suppose. I hadn't known that my friend and I shared a dentist in common, and I don't think anything nefarious will come of it. But this little episode shows that while some people concerned about computer security are worried now that the Obama Administration has gotten $19 billion out of Congress to spend on computerizing medical records, the old-fashioned paper records are not entirely secure either.

If you had asked someone in 1960 to guess whether doctors or lawyers would be faster to adopt computerized record-keeping, most people might have bet on the doctors. After all, doctors use advanced technology every day, while in 1960 it was still possible to operate a profitable law practice with manual typewriters and carbon paper. But history has proved this guess wrong. Most lawyers now shoot emails and .pdf files and electronic signatures around without a second thought, but doctors and hospitals still keep medical records more or less the same way they were kept in 1890: on millions of little scraps of paper in cardboard file folders. Yes, they can fax copies around, and sometimes do when requested, but the heart of the system is still paper, not electronic.

This state of affairs has its drawbacks. While no form of record-keeping is error-free, you would think that a profession with a reputation for bad handwriting would do something about keeping handwritten records of life-critical information before now. I am not aware of any formal estimate for how many people in the U. S. die every year due to medical errors caused by poor penmanship, but it's probably in the dozens, at least. So if it is done well, the transferring of medical records to computer form promises to reduce mistakes in a field where the phrase "fatal error" means more than just the fact that your computer crashed.

All the same, there are political groups which have been campaigning for increased protection of the privacy of patient records now that computerizing records looks like it actually may happen. A former RN named Deborah Peel runs an organization called Patient Privacy Rights, which tries to influence legislation to increase the formerly meager protections that U. S. citizens have against unauthorized use of their medical records. They have recently announced that the part of the stimulus bill paying for medical-record computerizing also has reasonably good protections in place for patient privacy. I admit to somewhat mixed feelings on this score, since efforts like this were responsible for the infamous information and consent forms every new patient has to fill out nowadays. To that extent, the reforms have increased everyone's paperwork burden, and whether patient privacy really got better is somewhat of an open question. But if the move to computers really succeeds, maybe the forms will become electronic too.

The $19-billion question right now is: exactly how is that money going to be used to convert the head-high file cabinets in every doctor's office and hospital in the country, into some computer files that presumably can be shared effortlessly from office to office? In trying to think of analogies to this, the only thing that comes to mind is the giant databases that aircraft manufacturers like Boeing maintain on new airplane designs. Everyone in the organization with the need to know about or alter these databases can access them instantaneously, and everyone ends up working from the same database. This system has increased productivity tremendously in the industry and reduced errors and misunderstandings a great deal.

But there are many differences between that situation and the case of medical records. Boeing is one private company, although a large one; the U. S. medical establishment is (so far, anyway) largely private, incredibly fragmented, and has no overall coordinating managerial structure of any kind. For the shared advantages of a common record-keeping system to be realized, somebody is going to have to impose some rules, which will include software and possibly hardware specifications. You know that all sorts of computer and software companies are currently salivating over that $19 billion, hoping that their particular product becomes the de-facto (or perhaps even de-jure) norm and they will end up with a good part of it in their pockets.

But given the rather anti-business tone of this administration, I can imagine another extreme, namely a Federal Medical-Records Computerization Agency, with thousands of newly hired young agents fresh out of college (where else are college graduates going to find jobs these days?), going from dentists to hospitals to chiropractors, knocking on doors and saying, "Hi, I'm from the government and I'm here to help you . . . computerize your medical records." Finish the joke any way you like, but that picture has its own drawbacks, not the least of which is the tremendous overhead in terms of federal employees and time it would take to decide on the system, train the agents to do their job, and then go and do it. If the Federal Aviation Administration's experiences with software are any guide (they have had tons of problems, including one or two expensive and complete failures), we have a right to be dubious that an all-government attempt to take on a huge software-intensive job like that will succeed, even with nineteen billion dollars behind it.

The sensible thing (and there are sensible people in government, though they don't often capture headlines) is an approach engaging both public and private entities, creating a minimum of new bureaucracies, trying things on a small scale first to work out the bugs, and adapting good systems that have been proven to work already. I hear that the Veterans' Administration, for example, has a pretty good computerized record approach that could be scaled up. Let's hope that sensible heads prevail, and we end up with computerized medical records that do the job with fewer errors and better privacy than the current paper records.

Sources: A report on some details of the portion of the stimulus package dealing with computerized medical records can be found at http://www.kansascity.com/444/story/1033620.html. Deborah Peel's Patient Privacy Rights organization has its website at http://www.patientprivacyrights.org.

Monday, February 16, 2009

The Crash of Flight 3407: Better Deicing Needed?

The passengers on Continental Airlines Flight 3407 from Newark, N. J. to Buffalo last Thursday night probably felt pretty confident about air travel, especially after hearing about Chesley Sullenberger's successful splash landing of his U. S. Air flight in the Hudson River less than a month ago after birds apparently clogged both engines. But a few minutes before the planned landing in Buffalo, the crew radioed that they were encountering noticeable icing on the wings. As soon as they attempted to lower the flaps for landing, the Bombardier Q400 commuter turboprop began to pitch and roll violently. The extended landing gear was retracted just before the plane crashed flat on top of a house in Clarence, New York, killing all 48 passengers and crew as well as one person on the ground.

Ice has been a problem for aircraft ever since the first airplane flew high and long enough to accumulate freezing rain on the wings. It tends to build up on the leading edges of airfoils. Besides its additional weight, ice can distort the airflow around the wing in unhelpful ways and even interfere with the mechanical movement of control surfaces such as ailerons and flaps.

There are three main approaches to deicing of aircraft. First, many airports are equipped to spray deicing solutions (basically a kind of antifreeze) on a plane's wings before takeoff. This can prevent ice buildup soon after takeoff, but it eventually wears off. In addition, larger aircraft use heated deicing strips that warm critical surfaces so that ice is less likely to form. Finally, smaller aircraft such as the two-engine Q400 usually use pneumatic deicing boots. If you imagine partially blowing up a balloon, spraying freezing water on it until it has a coating of ice, and then blowing it up more until it cracks out of its ice shell, you have pictured the essentials of a deicing boot.

At a news conference after the crash, federal investigators said that icing was a possible cause. The performance of pneumatic deicing boots has been of enough interest to inspire the Federal Aviation Administration to commission a lengthy investigation into how well they work. In 2006, the investigators published a 160-page report, which I have not had time to read since I found it this morning. However, the executive summary points out that in an actual flight test of an instrumented aircraft using the boots, enough so-called "intercycle ice" built up to produce a loss in lift of 25% or more. This loss became worse when the aircraft was close to a stall angle of attack.

The translation from aerodynamic engineering speak into layman's language goes something like this. Lift, the force that keeps an airplane in the air, can be reduced as much as one-fourth by ice that builds up between the times that pneumatic deicers operate (they work on an automatic cycle once they're turned on). If the airplane is trying to climb rapidly (as may well have happened, since a decision to retract landing gear is consistent with deciding to miss an approach and gain altitude), every airplane eventually reaches a "stall angle." When an airplane stalls, it nearly stops in the air and literally falls out of the sky. Recovering from a stall is hard, even in clear air in the daytime with a normally functioning aircraft when you have plenty of altitude left. The pilots of Flight 3407 were flying what was probably a plane with heavily iced control surfaces, in the dark, in freezing rain, with only about 2,000 feet between them and the ground. And it wasn't enough.

Speaking as a non-mechanical engineer, I have to confess that the picture of little rubber boots inflating and de-inflating to knock ice off a wing has its weirder aspects. Ice is not very strong mechanically, but it can be pretty sticky. And rubber exposed to the kind of life led by the leading edge of an aircraft wing is bound to get roughened and porous sooner or later, which will make ice stick to it that much better. I don't know any details such as how much the boots really inflate to blow off the ice. But the whole situation seems like it's a kind of empirical solution to a problem that is very hard to analyze theoretically, or even to model in the laboratory, as the FAA report itself admits. While the last accident where ice was implicated in the crash of this type of aircraft occurred fifteen years ago, even one life lost that could be prevented is too many.

We will have to wait for the final report of the crash investigation before we can draw any substantiated conclusions about what this accident says with regard to pneumatic deicing gear. As with many crashes, there may be a human factor involved. According to the Wikipedia website article "deicing," pilots used to believe that pneumatic deicers did not work well unless you allowed a certain minimum ice buildup to occur. Later studies reportedly revealed that this idea was false. We don't know what the late pilots of Flight 3407 believed, but we do have a record in the flight recording of what they did. Examination of that record plus engineering studies may show in more detail what went wrong and how accidents like this can be prevented in the future.

Sources: I used the MSNBC report from Associated Press currently available at http://www.msnbc.msn.com/id/29173163/, and the Wikipedia article "deicing." The FAA report "Investigation of Performance of Deicing Boots, Surface Ice Detectors, and Scaling of Intercycle Ice" is available for free download at http://www.tc.faa.gov/its/worldpac/techrpt/ar06-48.pdf.

Monday, February 09, 2009

Analog TV in the U. S.: Not Quite Dead Yet

Politics gets into everything nowadays, even my lecture notes for an electromagnetics course I hope to teach next fall. In the process of trying to give a more vivid picture of what the electromagnetic spectrum is like, I included a list of radio and television frequencies and what they were used for. One of them is "Old TV channel 56—now reassigned to other uses." When I wrote that, I was counting on having the switch to digital TV in the U. S. to happen on schedule, which until the politicians got into the act again was a week from tomorrow. But now, if President Obama signs legislation passed by Congress, which he has promised to do, the switch is delayed at least until June, and there is a good chance that even then, Congress and the President will take a look at the situation and say, "Well, there's still all these poor people who still haven't got digital converters, so let's wait another little while."

I confess to feelings of great ambivalence about the whole thing: digital versus analog TV, how the switchover has been handled here, and for that matter, TV in general. In these feelings, I continue a long-standing if obscure tradition of TV engineers who at best viewed TV in a dim light (metaphorically speaking) or disapproved of it altogether. The engineer who could be considered as the father of electronic television, Vladimir K. Zworykin, is quoted in the Wikipedia article about him as saying, "I hate what they've done to my child . . . I would never let my own children watch it." This feeling was shared by at least one other electronics engineer of note, Harold A. Wheeler, whose inventions were fundamental to both radio and television. For many years, my wife and I did not own a television set, and when we eventually got one, all we did with it was watch old movies on VCR tapes and DVDs. Finally, a couple of years ago when my father-in-law moved in with us, he received a large-screen TV as a gift and we hooked it up to cable TV. But I still suspect that in some ways the world might be better off without TV.

That being said, I earned my living for a couple of years around 1980 devising ways to keep people who hadn't paid for cable TV from watching it. This involved work at Scientific-Atlanta in what was called "scrambling." In the process, I had to get thoroughly familiar with the analog TV system, and purely for that reason I will be sorry to see the departure of an old acquaintance of long standing. Like any old friend, its flaws became as obvious to me as its virtues. In contrast to the European systems of analog color TV, the U. S. system (termed "NTSC" for the National Television Standards Committee that devised it in the early 1950s) did not reproduce the hue of colors very well unless the entire transmission system was carefully and repeatedly calibrated to maintain something called "constant group delay." For this reason, engineers joked that NTSC actually stood for "Never Twice the Same Color." Nevertheless, it served us reasonably well for over half a century, and I am somewhat sorry to see it go. If in fact it goes at all.

As to the ethics of the thing, I don't believe any wrongdoing can be laid at the feet of the engineers who cooked up digital TV and did the intensely political work of getting major companies and industry groups to agree on the new standards. By most engineering criteria, digital TV is a great advance over analog TV. It does more: it allows several sub-channels within one main channel, allows for a variety of display formats, and delivers an essentially studio-quality image everywhere within range of the transmitter, without the old analog problems of "snow" and "ghosts." (Of course, it does tend to fall off a cliff into complete disaster when you get out of range, but that is how all digital media tend to work: either very well or not at all.) And once it is implemented, we can get by with less spectrum bandwidth devoted to TV, which is how the federal government raised nearly $20 billion by auctioning off the surplus spectrum that will be freed up when all the stations go digital. This will lead, one hopes, to a variety of new digital wireless services, which is why some companies such as Qualcomm who were about to roll out such services complained loudly at the delay.

In comparison with how the digital rollout has been handled in England, the FCC and the TV industry stumbled rather badly here. In England, where everybody who watches TV pays a yearly license fee to the government to support the BBC and so on, the conversion happened in geographic stages and abundant help and equipment was made available. In the U. S., by contrast, there's been a lot of one-way communication in the form of advertising, an underfunded coupon program that assumes a lot of abilities on the part of the consumer (how to get the coupon, what to buy with it, how to hook up the box), and by some estimates, a failure to reach millions of people who still watch TV over the air with old analog sets. These folks tend to have lower incomes and are older and often socially isolated—people for whom TV is arguably one of their main companions. And I agree it would be a shame to simply cut them off without any help or assistance about what to do when that occurs. But no matter what we do, there will be some people who will be caught unawares no matter how much spade work is done in preparing the public for the change. And the rest of the country should not be held hostage to a few people who will have their screens go blank when digital comes along.

So maybe one delay is understandable. But such things can get habit-forming, like resolutions to quit smoking. Mark Twain said it was easy to quit smoking; he'd done it dozens of times. Let's hope we use the next four months to help more of the people who need help with the digital transition, and then go ahead and get it over with.

Sources: A news article describing the latest news on legislation to delay the switchover can be found at the Reuters website http://www.reuters.com/article/industryNews/idUSTRE5136YP20090205.

Monday, February 02, 2009

Engineering Ethics and "Software Engineering"

Every now and then I like to write about something that is at least potentially controversial. The question of whether software engineering is really engineering ought to do it. I'd like to be more specific: should people who call themselves software engineers be bound by the same professional ethical principles that other engineers claim to follow?

As types of engineering go, software engineering is a relative newcomer. Philosopher Michael Davis, who has written extensively on engineering ethics, traces the first use of the term to a 1967 NATO meeting on software design. Since then, computers and the software they all have to run on have become a huge part of everyday life, and an even greater part of engineering. There are seventeen accredited undergraduate programs at U. S. universities and colleges in software engineering, and by that and other measures you might think software engineers have as much right to call themselves engineers as any other member of the profession. But Davis isn't so sure.

That may be one reason that Davis, along with twenty-four other experts, contributed to the creation of a distinct ethical code for software engineers. It is promulgated by the Association for Computing Machinery (ACM) in cooperation with the Institute of Electrical and Electronics Engineers (IEEE). The IEEE Code of Ethics, which has been around for at least thirty years, is only 256 words long. By contrast, the full version of the ACM/IEEE Software Engineering Code of Ethics is over 2400 words long (although a shorter version is also available). More important than such superficialities as the length of the codes of ethics is the question of why software engineers need a separate set of ethical principles in the first place.

One reason may be that the education and training to do software engineering is markedly different than the typical training that other kinds of engineers receive. If you look at the undergraduate curriculum of most engineering programs, you see a solid one- to two-year foundation in the sciences: physics, mathematics, and (usually) chemistry. But it is generally accepted that people who can do good programming don't need to know any physics or chemistry, and even the utility of the kind of mathematics most engineering programs emphasize (that is, calculus, differential equations, and so on) is questionable. The type of science called computer science obviously relies on mathematics, but people without any significant background in computer science do software engineering all the time.

Are the ethical issues faced by software engineers markedly different compared to those faced by other engineers? The people who came up with the ACM/IEEE software engineering code seemed to think so, or else they would have simply referred inquirers to another code of ethics such as the IEEE's. A cursory reading of the ACM/IEEE code's long form reveals only a few items that could not explicitly apply to other kinds of engineers as well. For example, item 5.03 of the ACM/IEEE code states that those managing or leading software engineers should "[e]nsure that software engineers know the employer's policies and procedures for protecting passwords, files and information that is confidential to the employer or confidential to others." This is good advice to any type of manager, not just managers of software engineers. My sense is that, rather than leave some ethical stones unturned, the writers of the ACM/IEEE software engineering code tried to think of nearly every issue that software engineers might face, whether or not it pertains peculiarly to software engineering.

As a member of an older engineering discipline (electronic engineering), I confess to a twinge of professional jealousy as software engineering gains prominence. The truth of the matter is that as time goes on, the old divisions between disciplines become harder and harder to find in a typical workplace. It has always been true that many engineers also do management at various times, and often become full-time managers later in their careers. But nowadays it is hard to find any kind of engineer who doesn't at least use software, and every engineering student takes at least a smattering of computer-code writing along the way to graduation.

Still, there is the old notion that engineering is fundamentally about physical stuff, not the ephemeral and fundamentally non-material thing called software. Be that as it may, it is a hard fact that software is (a) produced by people with special knowledge for (b) use by non-specialists who (c) can be seriously inconvenienced (or worse) by software that doesn't perform as expected. Those three items have been true of all engineered products since we began to talk about engineers in the nineteenth century, and they are also true of the non-material product called software. So from a pragmatic standpoint, those who write software for sale or use by others bear the exact same type of responsibility as engineers who design bridges or rockets. For that matter, no bridge or rocket is designed today without at least the use of software, so by implication, software engineers are involved in most other kinds of engineering too.

Software engineering is still a young field, and news items about grand software-project disasters still come up from time to time. But the same was true of the earliest iron and steel bridges: they collapsed with alarming frequency. However, their designers didn't give up on the idea. Instead, they studied what went wrong, learned from their mistakes, got more organized as a profession, and went on to improve the next generation of bridges. I hope that the ACM/IEEE code of software engineering ethics does the same for its young discipline. But all the same, I'll take 256 words over 2400 words any day.

Sources: More thoughts on whether software engineering is really engineering can be found in chapter 3, "Are 'Software Engineers' Engineers?" of Michael Davis's book Thinking Like an Engineer: Studies in the Ethics of a Profession (New York: Oxford Univ. Press, 1998). The ACM/IEEE Software Engineering Code of Ethics is currently at http://www.acm.org/about/se-code. The IEEE Code of Ethics is at http://www.ieee.org/portal/pages/iportals/aboutus/ethics/code.html.