Co-host Shahid Shah attended the IBM World of Watson conference in Las Vegas this week. He was very impressed by the show IBM put on. He was even more impressed by the real-world demos he saw from organizations who are already using Watson to impact their healthcare businesses. In this episode, we discuss the current state of analytics, the fundamental shift that’s occurring thanks to machine learning and the role that IBM Watson is playing in that shift. We dig into the opportunities that exist today for everyone from large academic medical institutions to health systems to startups and other organizations launching new digital health products.

The state of analytics and the role of machine learning

Originally, analytics referred to programmers and data scientists crunching data and building reports that would be consumed by others and aid in their decision making. Next, Business Intelligence (BI) tools emerged and allowed non-programmers to pull their own reports on the fly from multiple data systems and build dashboards. Next, we evolved to predictive analytics where we could look at what’s happened in the past and use it to predict what might happen next. These models are based on algorithms that were created by humans to take an input, process it and generate a useable output.

Machine learning and cognitive computing are the next steps on this analytics continuum and they represent a fundamental shift in how we’ll use, and create knowledge. Instead of building the algorithms, humans are building the learning systems, training them and allowing them to generate their own algorithms. It’s a lot like teaching a child. Rather than strictly determining what the machine will do with the input, we’re giving it some guidance and boundaries and letting it draw its own conclusions. This may sound like science fiction, but it’s starting to happen today.

IBM Watson isn’t replacing doctors

The technology and its use cases are still young. We’re at least 5-10 years away from the point where IBM Watson and other cognitive computing platforms can begin to displace humans in any activities that require real reasoning and meaningful interaction with people. IBM’s aim in healthcare is not to replace doctors, but rather to assist them in delivering the best case they can. No human can read and internalize every healthcare study, treatment or practice that becomes available. Watson can. And Watson can utilize all of that knowledge consistently on every case it examines. It doesn’t get tired or have bad days and miss things. No matter how good they are, people do.

In the keynote speech at IBM WOW, they noted a cancer facility where Watson analyzed historical diagnosis and it found 99% of what it’s human counterparts did. What’s really interesting is that, in 30% of the cases, Watson also found something new. These are potential treatment options that the real doctors can now consider.

Watson can consume and use knowledge at rates that no human can touch. Doctors can reason and have meaningful interactions with patients Put them together and you have a powerful cancer-fighting team.

We get into this and much more on the show, including (use the time links to jump to that specific topic in the video):

  • The state of healthcare analytics and how machine learning changes it (3:30)
  • The IBM Watson / Doctor team (not replacement) (5:59)
  • Is IBM Watson building a single growing knowledge base for cancer research and diagnosis? (7:48)
  • How can we facilitate the sharing of expertise and intellectual property in knowledge systems in an ethical and financially equitable way? (10:47)
  • Who has an opportunity to get value from IBM Watson today? (15:49)
  • How are the health insurers experimenting with IBM Watson and other machine learning? (20:17)
  • What’s the difference between explicit data (i.e. data we have) and implicit data (i.e. data we can derive through machine learning)? What does that mean for business? (21:27)
  • How is machine learning impacting image and video analysis? (23:20)
  • This isn’t magic. How are we training computers to understand the knowledge they possess and apply it? (24:50)
  • How can we create scalable feedback loops to train cognitive computing systems when the volume of data is large and the human experts who understand it are few? We discuss this generally and in the context of internet security. (25:50)
  • What use cases are live in the field today? (32:50)
  • Will IBM Watson transform radiology? (40:48)
  • What opportunities exist for startups and other organizations launching new digital health products? (43:55)