Preventive medicine has always been about rules of thumb broadly applied. The rules tell us that these women should probably have mammograms, and those people probably need implanted defibrillators. But we're painting with a brush that's a mile wide. That's about to change.
Two examples illustrate how. One is the First Warning Systems sports bra which measures variations in skin temperature. Why? The company says that incipient tumors, as they begin to recruit blood supply, produce temperature signals that can be detected years before a mammogram could spot a problem.
The other example comes from MIT professor John Guttag. He studies electrocardiogram data looking for patterns that predict heart attacks. Why? Most implanted defibrillators, which are hugely expensive and quite risky to install, never activate. Meanwhile many people who ought to get them installed never do. Better prediction will lower cost and improve outcomes.
In both cases there are two distinct processes at work: data logging, and data analysis. And in both cases that data is going to wind up in the cloud. But whose cloud? By default it'll be the one used by the provider of the medical service. At first glance that seems to make sense. Isn't the company that invented the smart sports bra best qualified to analyze its data? Isn't the doctor who ordered your ECG best qualified to read it?
Not necessarily. The tight coupling between logging and analysis of medical data is just a historical accident. Your records used to accumulate in file cabinets at the hospital or clinic. Now they accumulate on hard drives there. But those drives are migrating to the cloud. That means we can rethink who controls the data and how it will be used.
Cloud computing is, after all, not only about roaming data. It's also about roaming computation. A decade ago we started talking about web services, and we identified loose coupling among web services as a best practice. Which it surely is, but not only for engineers of web services. It's also one of the principles of web thinking that we can all usefully apply.
When our data is loosely coupled to the services that analyze it, we can have a competitive market for algorithms. When you enter a relationship with a company like First Warning Systems you'll authorize them to access your temperature data and use their algorithm to analyze it. If you want a second opinion, you'll authorize another algorithm provider to take a look.
Or maybe you want to anonymize your ECG data and open source it so that students and entrepreneurs can use it to inform the development of new algorithms. Or maybe you want to adopt both of these strategies. They aren't mutually exclusive.
It's all possible. But it won't just happen. To make this future real you'll need to understand the principle of loosely-coupled services, envision yourself empowered with control of your health data, and vote for politicians who get why this matters.