Uncertain Health in an Insecure World - 45
Next week, I’ll Waze my way into lower Manhattan to join other healthcare leaders to interface with the IBM Watson Group. IBM is investing billions in artificial intelligence (AI) for business integration. Our university recently bought the suburban New York building where Watson was developed. Watson is a cognitive computer, made famous by its successful run on the TV game show Jeopardy. Unlike other big data crunching supercomputers, Watson’s advanced natural language processing capabilities helps it bridge the masses of text information in the medical literature and EHR patient records with structured genomics data.
The Cleveland Clinic, The Mayo Clinic, MD Anderson and Memorial Sloan-Kettering Cancer Centers are working with IBM on personalized medicine approaches, but the projects are lagging. That said, Harvard medicine professor and Beth Israel Hospital Chief Information Officer John Halamka harvested 3 petabytes of big data from all the Harvard hospitals in 2011, and used his wife's personal genomics analytics to help doctors cure her breast cancer. "In today's healthcare system, it takes, on average, 20 years for an innovation from one hospital to diffuse throughout the country", worried Dr. Halamka. In his CXO-TALK (@cxotalk and http;//www.cxotalk.com), he also warned about all the big data applications being hyperbole from vendors. But at a personal care level, for his wife, big data analytics was a game changer!
I am not a computer scientist, so I've decided to study up before meeting Watson. Of course, because all the world's computer scientists, Hadoop programmers, and algorithm diviners are populating my Twitter feed, I went directly to the source. At the IBM blog site, I read about "enabling the world: IBM Bluemix, the cloud and cool apps". Apparently, to interact with Watson, you will need to "add and angular JS interface to a Node-RED workflow...". OK... sounds easy enough, and there is a diagram (see below).
Amid the explosion of personal Internet-of-Things little data and the streaming of unstructured medical device data, behind the scenes progress is being made by math and engineering 'quants' on tagging key data and building smarter algorithms. Quasi AI machine learning firms can easily iterate thousands of model algorithms weekly, helping Osco Pharmacy to figure out the people who bought beer when buying diapers. But only the quants know what goes into a model, and how to make sense of it.
Walmart and its 250 million weekly customers generate 40 petabytes of personal transaction data daily... "Walmart has made big data part of their DNA". Big Pharma also collects many petabytes of personal prescription information daily from pharmacies and patients using tens of thousands of sources – trying to leverage this big data to predict populations that might be receptive to new medicines. Conversely, in their clinical trials, Big Pharma companies use big data analytics to de-risk their investments in new drugs by discerning which of these drugs are failing fast, in part in order to conserve their R&D budgets.
Computer scientists have long dreamed of building machines that can mimic our minds and preserve our memories after death. Neuroscientists are coming around to the idea that brain-to-computer data transfer is not science fiction. When Ex Machina meets the cloud, the ending may not be a happy one.
What does ALL this data and computing capacity coming together really promise?
I plan to ask Watson the question that 'Joshua' was asked in War Games (1983)...
Until that meeting, we in the Square have a very human intuition that much of the future promise of AI is actually hyperbole in the present.