Back

 Industry News Details

 
Interview with John Langton, Director, Wolters Kluwer - Speaker at Global AI Conf - Sep 2018 Posted on : Sep 17 - 2018

We feature speakers at Global Artificial Intelligence Conference - 2018 Sep 25 - 27 – Boston to catch up and find out what he or she is working on now and what's coming next. This week we're talking to John Langton, Director, Wolters Kluwer Topic - "NLP For Medical Information"

Interview with John Langton

Tell us about yourself and your background.

I am currently the Director of Applied Data Science at Wolters Kluwer Health. I have a background in AI and hold a Ph.D. in computer science from Brandeis University. I started my career conducting research and development for the Department of Defense. I applied that experience in the private sector by founding my own company called VisiTrend. My startup focused on applying machine learning to cyber security data to identify threats and conduct forensics. VisiTrend was acquired by Carbon Black in 2015. Afterwards, I joined athenahealth to apply AI within the healthcare domain. It was this experience that ignited my passion for AI in clinical decision support which is a major part of my focus at Wolters Kluwer. While I do not publish as much as I did in my academic days, I still try to attend events and share research insights as I am able.

 What have you been working on recently?

 I support multiple initiative across Wolters Kluwer Health companies including but not limited to: a) machine learning algorithms to predict the onset of clostridium difficile infections with our Sentry 7 product from Pharmacy One Source (http://www.pharmacyonesource.com/), b) automated risk assessment of literature for pharmacovigilance using our Ovid product (http://www.ovid.com/), c) AI enabled readmission risk assessment from our EMMI population engagement data (https://www.emmisolutions.com/), and d) several NLP projects related to processing both well-formatted text like research literature and noisy text like doctors notes from electronic medical records.

 Tell me about the right tool you used recently to solve customer problem?

 We focus on innovation withAI that requires multiple tools along with custom implementation. I’m personally fond of the H2O platform for scalability but most of our work is done directly in python. Tools are rapidly being developed tosupport the full AI workflow but we’re not quite there yet. And many real-world problems will always require customization.

 Where are we now today in terms of the state of artificial intelligence, and where do you think we’ll go over the next five years?

 AI has become more accessible to folks who do not necessarily know how it works. I believe this trend will continue with tools that increasingly automate functions of an AI workflow. It is likely that businesses will mistakenly shift investments from high skilled workers to tools, but the pendulum will likely swing back once they see there’s a limit to what problems can be addressed without someone who actually knows what they’re doing. There is kind of a core set of algorithms that do pretty well on most problems, but I do think there will be incremental improvement on more complex algorithms. Some folks may start becoming more serious about addressing general problem solving again or what is sometimes called “strong AI”. But I think in 5 years we will primarily see continued advancement in weak AI. 

 There is a negative perception around AI and even some leading technology folks have come out against it or saying that it’s actually potentially harmful to society. Where are you coming down on those discussions? How do you explain this in a way that maybe has a more positive beneficial impact for society?

 AI is a powerful tool that enables both good and bad outcomes. Which outcome is realized depends on the humans using it. The reality is that the advancement of AI is inevitable. At Wolters Kluwer, we are leveraging AI not to replace doctors but to give them better information for making better decisions. We aim to revolutionize healthcare through AI and we are well on our way with our current projects.

 When you’re hiring, what types of people are you hiring? The job market for traditional programmers, engineers is very difficult to get into AI space. Are you hiring from that talent pool or is that a different talent pool? In terms of talent, how do you go about ensuring you get the best AI people at your company?

 I hire folks who have great quantitative and logical intuition. This typically entails having an advanced degree in a quantitative field, but not necessarily computer science. Indeed, I have hired folks with advanced degrees in physics, math, etc. Familiarity with computing and algorithms is essential, and experience with relevant algorithms is definitely a plus. But ultimately, I am not looking for people who know specific machine learning APIs, but for people who can learn. In fact, there is a glut of folks who can download a python library and call an API method but who are not able to figure out why something does or does not work. I look for folks who can get an intuition of the information space, how algorithms work, and think in creative ways to solve complex problems with AI.

 Will progress in AI and robotics take away the majority of jobs currently done by humans? Which jobs are most at risk?

 Many jobs will likely be automated but it is hard to say whether the majority of jobs will or will not be automated. Such decisions are based on several factors including cost, scalability, and complexity of deployment. In many cases, it may be more cost effective to have humans do a job than to deploy an expensive robot or an AI solution that must be maintained over time. The complexity of deployment and changing existing workflows may also be too costly to justify automation. However, people do make irrational decisions. Executives may overestimate human resource costs and undervalue human decision-making capabilities. The jobs most vulnerable probably include those with repetitive tasks. However, as AI advances, the types of jobs that can be automated may change.

 What can AI systems do now?

 This is a difficult question as it is not clear whether you are talking about practical use cases, computational limits, or benchmarks for intelligence. AI has been used in several areas of the healthcare domain but has probably had the most success in image processing and radiology. Predictive analytics are still not that great for alerting on disease conditions, but research is coming along rapidly.

 When will AI systems become more intelligent than people?

 AI systems think differently than people, so this question doesnot quite make sense to me. People are great “general problem solvers” which was the original goal of AI (thatit failed to accomplish). At some point, AI may overtake humans in terms of general problem solving. This is the holy grail of “strong AI”. I am skeptical as to whether that is possible with current approaches. If it is, I wouldgive it something like 50 years.

You’ve already hired Y number of  people approximately. What would be your pitch to folks out there to join your Organization? Why does your organization matter in the world?

 We are working on the most important problems such as curing disease and improving healthcare. There is no position where you could have a greater positive impact, not only with our customers, but with theirpatients, customers, and society as a whole.

 What are some of the best takeaways that the attendees can have from your talk?

 I can provide some context about how AI is used within healthcare.I can share one of the projects we are working on and hopefully a few tips/tricks for dealing with particular problems in applying AI to messy medical text data.

 Which company do you think is winning the global AI race?

 It will be interesting to see the “machine learning as a service” offerings of Amazon, Microsoft, and Google evolve. Other than that, H2O is the most effective large-scale AI platform I have used to date. Within healthcare, I think there are several folks doing interesting things. However, Wolters Kluwer is in a unique position to win Healthcare AI. We have access to critical medical data, subject matter expertise, and the most prized content around. Combining this with our growing team of AI experts means you will be seeing some very exciting developments coming out of our company.