There’s a saying that goes, “Everyone who thinks they’re saying something important has the same chart. It looks like a hockey stick, and the lines are shooting up right about now.” This is exactly what is happening right now in healthcare Big Data. At the Medical Innovation Summit yesterday, I was able to listen to Dr. Eric Brown, the Director of Watson Technologies at IBM, who he gave a fascinating talk on the future of artificial intelligence, Big Data and healthcare. The evolution of computers will push healthcare forward, and healthcare administrators from every discipline should be thinking about how we can use Watson-like computers for better outcomes.
First, let’s start with the basics: artificial intelligence programs like Watson need to be able to understand a) complex language and b) find the answer across a huge range of subjects—in fact, really all of human knowledge. Second, you have to understand the way we think about trivia questions, or data in general, which often involves finding an intermediate answer that gets you to the final answer. To build Watson, IBM built on previous research in natural language processing, machine learning, parallel and distributed computing, and much more—all with an incredibly low latency that gave the answers in a Jeopardy-fast 3 seconds.
As we move into 2014, we’re very quickly moving towards the era of “cognitive systems,” which will help us make meaning of vast universes of unstructured data. Moreover, these systems will learn and evolve over time, helping humans find patterns and insights that we never would have otherwise. The computer will not only answer the question, but it will help refine it, converse with the user, and then offer evidence and explain why it chose that particular answer. In the case of healthcare, one Cleveland Clinic doctor mentioned that he has 1,000 pages of text information on each patient … and in some cases, 15,000! Not to fear: Watson can handle all of this easily and may eventually be able to integrate radiology images and even the patient’s genome into its answers. Dr. Brown showed a case where Watson took some basic doctor notes and suggested Parkinson’s disease as a possible cause. In another example, Watson took a patient’s symptoms and urinalysis into account when diagnosing a patient—all via an iPad.
For the healthcare marketer, assuming he/she has a robust database of marketing activities and their results, artificial intelligence will be able to deliver much more than just a single ROI figure. Conceivably, a Watson-like computer will be able to leverage your EMR and other databases to dictate the specific message that will work with a specific population at a very specific time. Of course, one major challenge will be reporting: How can Watson present information—clinical, marketing, or otherwise—in a way that humans can understand? Will Watson get so good that it replaces clinical providers, or at least makes them redundant?
All of these questions just make healthcare technology the most exciting industry on the planet. Do you agree? Let me know in the comments below.