I read this story wanting to understand if the data mining they’re doing is really appropriate for making individualized statements in the way they are claiming when they suggest that hospitals will get risk assessments based on patient shopping data through credit cards, store cards, etc. Will receiving doctors get sufficient training in the ways in which these predictions are like and unlike the predictions that medical tests make about health risks?
Additionally, I read through the list of hypothesized triggers for heath risks and they seem to bank on the idea that everything I’m purchasing is for myself. Just in the first paragraph the article suggets issues if “ou’ve let your gym membership lapse, made a habit of picking up candy bars at the check-out counter or begin shopping at plus-sized stores” which could nicely match my patterns this year of regularly picking up candy for the lab and buying some clothes for an elderly relative who can’t get out to stores as easily anymore. The majority of the time I buy donuts, I do not actually eat any of the donuts – and I have colleagues whose donut-eating patterns may more closely match my shopping trends than theirs.
And then you get this statement: “While the hospital can share a patient’s risk assessment with their doctor, they aren’t allowed to disclose details of the data, such as specific transactions by an individual, under the hospital’s contract with its data provider.”
I’m not sure if that’s good — hooray for not sharing personal details! — or worse — so the computer says I’m at risk but we can’t sit down and talk about whether the patterns it’s identifying are real. And, going back to my opening question – what sorts of algorithms are being used and given that, what sorts of conclusions are even valid to draw.