It’s always possible to have too much of a good thing. But too much innovation?
At the risk of aiding and abetting those who believe in the end of software innovation — something I fundamentally disagree with — I have to admit I’ve recently seen another compelling reason why we need to temper our views of progress with a little bit of reality. Or pay the price.
I ran into a similar problem in the late 1990s, when Internet net markets were being sold to top-tier manufacturers as a way to further squeeze out their second- and third-tier suppliers, who also were being asked to pony up millions of dollars in order to connect to the very net markets that were designed to further undercut their margins. It was a classic case of doing something largely because the technology can, no matter the consequences. Those net markets failed, in part because the suppliers basically refused to play ball.
The latest example of “good technology, bad business model,” may, unfortunately, succeed.
My new candidate for “too much innovation” is MEDecision, a medical management vendor selling largely to HMOs. MEDecisions raison-d’etre is to provide HMOs with software that lets them better manage patients and doctors, with a wishful eye toward providing better service to patients. They do this by compiling, or trying to compile, tons of data about patients, their diagnoses, treatments, and lab tests, and then using that data to run a system that provides automatic responses to authorization requests for further treatments, lab tests, and medications.
That’s only the beginning of the MEDecision business plan.
The end game is that this system will eventually offer alternative treatment options to doctors based on an analysis of patient data. The result, according to MEDecision, will be a whole lot of cost savings and a whole lot more time for doctors to spend with their patients instead of monkeying around with those pesky authorizations, or even, in the end, the diagnosis process itself.
Among the many things wrong with this model is the data access and quality issue. Automated decision-making systems are as smart, or as stupid, as the rules they work with and the data they process. When it comes to Amazon.com recommending books to you based on past buying history, rules and data are pretty much in synch: they know enough about your buying habits and the buying habits of their other customers to accurately infer, at relatively low risk, what other books you might want to buy.
But feed a rules-based system bad or incomplete data, and the garbage starts to flow.
The problem is that complete data is going to be hard for MEDecision to find. Medical records are still largely stuck in the pre-Gutenberg era, hand-scratched on paper and encoded in cryptic acronyms. Furthermore, most HMOs don’t even own this data. It’s the property of the patient and the doctor. So in order for these automated decision-making systems to work, the HMOs need to force the doctors and hospitals to pony up the data — at their cost — and then hope that it arrives both clean enough and complete enough to do the job.
And if the data is bad, guess what happens?
At Amazon, you’re offered a biography of Barry Manilow when you’re really trying to read the complete works of Dave Barry. No harm there. But if you’re a medical patient with a complex diagnosis and an incomplete medical record in the MEDecision system, it’s potentially quite harmful.
Of course, in many cases, authorization denied is a relatively benign problem. Unless it happens on a Friday afternoon, or a holiday, or your doctor is out of town, or your meds have been stolen along with your purse, or you’re sick and the system is dead wrong and there’s nothing you can do about it. The combination of rigid rules, poor data, and problems with creating flexible exception handling make it hard to be excited about further automating authorizations the MEDecision way.
Unfortunately, MEDecision has the right idea, but the realities of data and healthcare make it impossible to do the idea justice. It’s like the net market idea: Just because we can doesn’t always mean we should. Sometimes too much innovation is just that: too much.