News Stay informed about the latest enterprise technology news and product updates.

IBM discusses plans for the Watson supercomputer, predictive analytics

IBM vice president Deepak Advani talks about the company’s SPSS acquisition, predictive analytics strategies and the future of the "Jeopardy!"-winning Watson system.

Image goes here

The world was fascinated last month when IBM's Watson supercomputer -- the result of four years of artificial intelligence and analytics technology research and development -- took down legendary Jeopardy! champions Ken Jennings and Brad Rutter. But Watson's appearance on the popular trivia game show was only the beginning. According to IBM, versions of Watson will one day be used to help doctors quickly diagnose diseases, help financial organizations detect fraud, and a whole lot more. got on the phone recently with Deepak Advani, IBM's vice president of predictive analytics, to find out how Big Blue's predictive analytics program has been progressing since it purchased predictive analytics software maker SPSS back in 2009. But the conversation quickly turned into a discussion about IBM's future plans for Watson. Advani also talked about how one California-based hospital is using data mining and analytics to save lives, and he had some advice for organizations considering a predictive analytics program.

What are the main drivers behind IBM's investments in SPSS and predictive analytics?

Deepak Advani: [Research firm IDC is] saying in the last five years, the amount of digital data that has been created has gone up 10 times and over the next 10 years, it's going to go up 29-fold. These are huge numbers, right? And a lot of this data that is getting generated is unstructured data, [about] 80%. There are blogs, social media, call center records, emails and whatnot [and] you want to take all this data, analyze it and then hopefully predict what is likely to happen, and that takes hardware software and services, and that is where we've been putting our bets. One of the great examples of that is what we did with Watson and the Jeopardy! challenge. The ability to digest a million books and, using natural language processing, to respond back within three seconds -- it's something we're very proud of. [We're] taking this technology and applying it to business problems, [and] we believe [this] is going to change the game and it's going to be very revolutionary.

Human beings may not always be rational, but they are very often predictable.

Deepak Advani, vice president of predictive analytics, IBM

Other than suggesting that people rush right out and purchase SPSS software, what advice do you have for organizations interested in predictive analytics?

Advani: The advice I would give to most people is first start paying attention to what people are saying about your brand. Go to Google Trends and subscribe to Twitter feeds on your brand. It doesn't cost you anything. At least, you'll then be in listening mode and aware of what people are saying. That is step one. And I think that what people will realize is once you start doing it, a whole world of conversations open up that they're not engaging in. And slowly, [you'll] start to say, ‘OK, well, that's interesting. What do I do next? How do I start to engage in this conversation? How do I get insights from all of this data so I can start building propensity models and start predicting what is likely to just fade away and what is likely to become a huge issue for my company?’

How is IBM customer Sequoia Hospital making use of SPSS and predictive analytics technology?

Advani: That is one of my favorite examples because at the end of the day what inspires all of us making this world a better place. Sequoia has been using SPSS and predictive analytics for a while, for many years, and they focus on cardio surgery and they've been able to reduce their mortality rates by 50%. They've been the No. 1 hospital in the country for valve replacement surgeries. The way they go about it is they've collected patient data for 10,000 patients in their database, and when somebody new comes in, the software, using data mining will tell you: 'Wait a minute; this new patient has got diabetes [and his] hemoglobin is elevated, so don't go into cardiac surgery. Do some different preoperative procedure.' [They're using predictive analytics to] reduce risk and do a better job of diagnosing and they've seen better business outcomes.

Could a version of the Watson supercomputer be put to work in hospitals one day?

Advani: If you can start looking at what we're doing with Watson -- a million books, 200 million pages, have all been digested. An average physician spends five hours a month keeping up on technical journals and all of the reports that keep coming out because they're just inundated with so much information. But if you have a system like Watson that is digesting all of this unstructured information and you have a Q/A system that now is providing insights coming from millions and millions of pages of information for a physician, you can start to see how that would improve health care even further. [For example], you could have Watson for oncology or Watson for pediatricians, and it could say that based on the latest research coming out of Harvard, pay attention to this [particular] symptom or another.

How does the Watson supercomputer fit into IBM's overall plan for the future?

Advani: The Watson machine is not connected to the Web, so everything that it needs to answer, all of these questions are all within the hard drives and within the systems that you saw on TV and that you hear about. We are now in the process of commercializing Watson, and it's going to happen in a couple of different ways. I'm just sharing our latest thinking because we don't have a product in the market today. But the way you can envision it coming to market is you can see a lot of the Watson insights being delivered using Software as a Service, through the cloud. You could envision a scenario where you have a smaller version of Watson in an appliance that is customized for health care or customized for fraud detection. You can see scenarios where you have a hybrid of the two. We are in the process of looking at the applicability of the Watson algorithms industry by industry, business process by business process, to see where we can add the most value.

How about plugging Watson into the Internet and taking on Google with a Watson-powered search or question-and-answer engine?

Advani: IBM is all about delivering business outcomes or appliances industry by industry. So, our focus is much more on how can we take a lot of the algorithms and inventions we've done with Watson and connect it to all the other assets that we've built up over the last decade or so and solve problems for clients. A lot of our focus is looking at fraud detection, reducing risk, improving health care, improving crime prevention, and we're looking at it through the industry lens.

Dig Deeper on Predictive analytics

Start the conversation

Send me notifications when other members comment.

Please create a username to comment.