Saturday, September 22, 2007

Artificial Intelligence

A funny thing happened to me on the way to the doctor’s office. There I was, reading a computer magazine’s reviews of various voice recognition products when I suddenly realized that all of these reviews were written by extremely computer-literate journalists. Each review was then polished by professional editors before the text appeared in print. Because of their superior language skills, these people had a much different experience with voice recognition software than John Q. Public might have. Voice recognition still faces severe challenges. I listen to some of them every day. They’re called doctors. And if you think I’m kidding, then examine some examples of their work. Keep in mind that this is exactly what these physicians said while dictating:

  • "Impression colon common migraine headache plan colon patient to common migraine headache patient to plan colon patient to try Midrin and Fioricet on a p.r.n. basis.”

  • “I discussed with the patient that he should find a habit so he can do any exercise and have some mental for relying so that he able to have some more or less regular lifestyle and in the meantime I will provide some Ambien 10 mg p.o. q.h.s. to see whether this will helpful or not.”

  • "Past history the usual up to the usual."

There is absolutely no question in my mind that voice recognition has great promise. But last year I went head to head with a rude and obnoxious surgeon who screamed “Why are you typing what I say instead of what I want?” The bottom line is that I own a medical transcription service, not the Psychic Friends Network. The fact that someone is a doctor does not impress me because -- even under duress --many doctors couldn’t edit their way out of a paper bag. Just because someone got through medical school does not mean that that person speaks fluent English.

So, before we get a little too intoxicated with the promise of new technology, perhaps we ought to pause for a good stiff reality check. Voice recognition depends on computers. You might think you’re improving the situation by taking the human factor out of the dictation/transcription process but when you totally rely on computers, you might actually be making things worse.

A computer’s cold and calculating objectivity can cause problems for a doctor’s ego. Some doctors love experimenting with new technologies. Others are extremely technophobic (I recently got a call from a doctor who wanted us to transcribe her work but admitted that she couldn’t take electronic delivery of documents because she refuses to get a computer for her office).

Some doctors can keep their minds focused on doing one thing at a time (which helps when you’re using voice recognition). Others are trying to do three things at the same time while fielding interruptions from support staff.

Some doctors create well-organized reports that have a beginning, a middle and an end. Others ramble on in stream-of-consciousness dictation, fully expecting that a secretary or medical transcriptionist will clean it up and make it sound right.

Some doctors make statements which they think are crystal clear but which make no sense at all! The following two examples are taken from dictation by a physician who used to be the chairman of his hospital’s Medical Records Committee.

  • “The patient is a 30 year old right handed married mother of one child age four who is a former office manager in a family business in steel fabrication and who was placed on permanent disability in February of 1997.”

  • "The patient is a 35 year old right handed woman who lives with a partner who has been pregnant on two occasions and who has had two abortions and who did take birth control pills which she discontinued about 10 years ago because of profuse bleeding. "

Misplaced modifiers, dangling participles, blazing contradictions and run-on sentences! These are the stock in trade of most dictating physicians. Add in foreign accents, incorrect lab values, rotated right/left designations, wrong genders and you start to encounter problems that voice recognition might not readily solve

While artificial intelligence can ask the computer to question potential contradictions, the truth of the matter is that the computer is an extremely democratic device. Whether it runs an ATM machine or a digital dictation system, it does not care about how much money you make or what kind of car you drive. The computer does not care whether you are white, black, Asian, Hispanic, male, female or a one-eyed albino transsexual from Greenland. Computers have absolutely no emotional attachment to the data they store and interpret. Artificial intelligence is just what it claims to be: Artificial.

Because the human factor is not part of voice recognition software, the computer doesn’t have to care. It doesn’t have to care about going back and putting in a medication that was unintelligible at the top half of the report (but which the transcriptionist recognized at the end of the report). It doesn’thave to care about stroking a surgeon’s ego to get a report done on time. It doesn’t have to care about going the extra distance in order to hold onto its job.

And because it doesn’t have to care, it won’t.

"So what does that mean?” my mother asks her doctor. “That the patient is dying of improvement?”

There are many ways that voice recognition can enhance the dictation/transcription process -- if it can overcome certain obstacles. But there are too many situations where we can jeopardize patient care by relying too heavily on the two forms of artificial intelligence positioned on opposite sides of the keyboard.

Look at the software: Voice recognition doesn’t care whether the person dictating is a falling-down drunk, a punked out 16-year-old with a 9-gauge steel ring through his nose, a porn star who wants to write poetry or a physician who drives a Lexus. This means that a physician cannot bully the software. Nor can he try to impress or intimidate a computer with the fact that he is a doctor

That’s a humbling thought. And one that doctors would be well advised to remember when they rush to embrace voice recognition technology after telling medical transcriptionists that “it is beneath a doctor’s professional dignity to have to enter numbers into a telephone keypad.” Or that doctors shouldn’t be expected to enter the patient’s medical record number while dictating because “that’s a clerical function.” The computer doesn’t give a rat’s ass about the fact that someone is a doctor. The computer wants data.

Now look at the humans (dictating physicians) who are supposedly going to want to use this software. These people take great pride in hiring others -- caterers, maids, accountants, financial planners, gardeners, interior decorators -- to do work for them. With that thought in mind, has anyone stopped to ask why someone becomes a doctor?

  • To treat sick people? Yes!

  • To get lots of ego gratification? Sure!

  • To make lots of money? Damn right!

  • To achieve social status? You betcha!


Does anyone become a doctor:

  • To become a data entry drone? Don’t think so!

  • To become a glorified secretary? No way, Jose!

  • To be glued to a computer like a temp worker? Gimme a break!

  • To decrease one’s hourly rate of compensation by at least $100? Fat chance!

As voice recognition technology makes deeper inroads into popular culture it is interesting to see where it works well.

  • If I call any major airline to check on a flight’s arrival time, I probably won’t have to talk to a real human being. Instead, I can receive inbound flight information through a sophisticated computer system that utilizes voice recognition and artificial intelligence technology.

  • I can now order my prescription drugs through Walgreens Pharmacy’s Touch-Tone Prefill service which, by using a series of computer-generated prompts, enables me to dictate information into different fields. Alas, this doesn’t always guarantee that my order will be filled on time.

How quickly will voice recognition replace medical transcriptionists? In certain situations, where there is a doctor who is nerdy enough to want to do his own dictation (and who can edit his own work), this might happen. In specific departments like radiology, pathology and the Emergency Room, it might happen. But if the pressures of the market come to bear, I think you’ll be in for a big surprise.

Let me explain.

I live near San Francisco’s Castro district, which is heavily populated by gay men. Thousands of them pass through three city blocks on their way to work every day. Most of them are extremely fashion conscious. Most have a higher level of disposable income. Most of these men own an iron and know how to use it. But in this tiny little neighborhood, there are at least six cleaners running specials on tailored shirts (two opened their doors for business this year alone)!

Why aren’t these people doing their own ironing? Because that’s not how they want to spend their time! And when it comes to doctors, I have a hunch that voice recognition is going to face a similar challenge.

This reminds me of an old joke about the couple that was vacationing at the Fountainbleu Hotel in Miami Beach. All week long they watched what had obviously become a poolside ritual. Each day, a particularly handsome young man was carried out to the pool by servants. Drinks and food were brought toh im on a regular basis and he was obviously the life of the poolside party. Curiously, though, he never got up from his chair. When it was time for dinner, servants arrived to carry the young man back to his hotel room. On the last day of their vacation, the visiting couple stopped to offer their sympathy to the young man’s wealthy parents.

We’ve seen your son get carried back and forth to the pool every day and it just seems like such a tragedy. He’s so young. So handsome. So full of life. And to think that he can’t walk!”

“What tragedy?” laughed the young man’s mother. “Of course, he can walk. But thank God, he doesn’t have to!”

With that image fresh in your mind, does anyone want to take bets on how easily voice recognition will replace professional medical transcriptionists?

No comments: