I just came back from a systems biology conference at the Sanger where I heard a talk by Keith Baggerly, well known as a practitioner of “forensic bioinformatics,” the dark art of using raw data and reported results to reconstruct what the methods must have been.
In a previous post on de-discovery I mentioned his debunking of fishy Duke breast cancer results and clinical trials. Here, statistics was actually key to saving human lives. All of Keith’s results –reproducible as they should be– are collected here.
His talk yesterday was an update on his talk last year at the CRI (at videolectures.net, where you can even see me running in and out of the picture in the beginning and end of the talk). Keith is a great speaker, check it out!
So, what’s new?
Well, surprisingly, the Duke story is still not dead. It’s going on for years now. And finally mainstream media –and not only the Annals of Applied Statistics— have caught on. The New York Times wrote:
And medical researchers see the story as a call to action. With such huge data sets and complicated analyses, researchers can no longer trust their hunches that a result does — or does not — make sense.
“Our intuition is pretty darn poor,” Dr. Baggerly said. *
Also, the length of Anil Potti’s list of retractions is growing longer and longer. And I heard, Keith will be on TV soon.
I’ve asked him to put me on his “spam list,” an email list with regular updates. Let’s see how the story evolves. Keith Baggerly said he was sure it’d be all over soon–but then again, he had already said that last year …
Image source: screenshot from Cisplatin vignette.