I recently returned from the TED conference -- an annual conference that, as they describe it, "brings together the world's most fascinating thinkers and doers…" The speakers at this conference both motivated and scared me. I walked away with a very similar feeling to Rick Smith. And I, like many other people, left the conference motivated to make a positive impact on the world.
But, I'm not likely to develop weapons against drug-resistant superbugs (like Bonnie Basser), to develop new surgical robots (like Catherine Mohr), or prevent pandemics before they happen like epidmiologist (like Nathan Wolfe). I'm just an Internet guy. What can I do?
It was when I heard Tim Berners Lee speak about Linked Data at TED that I had an idea. I've been blogging about the Implict Web for some time now. Specifically, I believe that as people conduct more of their life online, they leave a mountain of digital breadcrumbs (also called data exhaust). Historically, this data has existed in silos. Amazon knows what books I like. Netflix knows what movies I like. OpenTable knows where I eat. But as more website adapt web services and build APIs, this data can be used in new and unexpected ways. And I believe that tremendous value will be created when someone can" cross the silos". Google's PageRank, for example, is a perfect example of data exhaust. It's a set of data (webmasters HTML links) that was created for one purpose, but adds a ton of value when used in a new way (search).
So, here’s my latest hypothesis:
I believe that we're rapidly reaching a point where we can learn as much about disease causation by leveraging internet breadcrumbs as we can from traditional medical studies.
For example, there has been an ongoing controversy over whether the use of antiperspirants/deodorants increase the risk of breast cancer. While there have been several studies done to date, the results haven’t provided conclusive data. Indeed, the National Cancer Institute has said: “Because studies of antiperspirants and deodorants and breast cancer have provided conflicting results, additional research is needed to investigate this relationship and other factors that may be involved.” While I’m not a medical researcher or statistician, I’m amazed about the low sample sizes of these studies. For example, the most recent study cited by the NCI had just 104 people in it. The study before that had only 437 people in it.
What if we asked people to opt-in to participate in a Medical Data Pooling Project? Users would voluntarily contribute their data exhaust (without any personally identifying information) for medical research. Every year hundreds of thousands of people support the fight against breast cancer by walking/contributing to walkathons (the Susan G. Komen Walkathon, The Avon Walk for Breast Cancer). What if we asked them to join a linkathon -- just link your data to this nonprofit medical data pool?
For example, CVS has 50 Million people who have enrolled in Extracare (their loyalty program). I sure bet a lot of them are women who buy deodorant at CVS. And if you don’t shop at CVS, you probably buy deodorant at a supermarket that has a loyalty program (most likely run by Catalina Marketing). What if a researcher was able access this global consumption data and target a segment of the population (say, 50,000 women above the age of 40 who buy deodorant 4 times a year or more, for example) and then survey them about their medical history? The ability to link to consumption to medical condition would be extremely valuable. The power of this linkage could be immense.
Take the debate over whether the chemicals used in dry cleaners (PERC) cause cancer. With over 600,000 users tracking their expenses, I’m sure that Mint.com would be an invaluable data set for medical researchers. Want to study if prolonged cell phone usage causes cancer? Skydeck collects your cell phone usage data.
Convincing users to anonymously part with their data and answer some medical questions is not going to be easy...BUT, it should be a lot easier than convincing them to participate in clinical trials. Current non-profits could be enlisted to get their membership to participate (ie, it might be easier to get someone to donate their data exhaust than to give money or participate in a fundraiser). And I have to think that the database providers would love this -- after all, this would (1) potentially encourage more consumers to participate, (2) benefit society and (3) provide them with a compelling argument to all those people who oppose database marketing.
Anyone up to working on this with me? What am I missing?