Big Data Analytics for Healthcare: A Q&A with Lance Speck of Actian

Originally published 28 October 2014

This BeyeNETWORK article features Ron Powellís interview with Lance Speck, general manager of the healthcare division at Actian. Lance and Ron discuss how big data analytics is being used in todayís healthcare organizations.
I understand that you lead the Actian healthcare business. Can you tell us about your background and the solutions Actian offers for the healthcare industry?

Lance Speck:
Iím with Actian. Iím general manager of the healthcare group, and Iíve been in the data space for about 20 years. For about the past ten years, Iíve focused on healthcare and other verticals, but healthcare being one of the ones that has had the most recent spike in growth. Most of my focus right now is in big data and analytics related growth in the healthcare industry.

What challenges are healthcare organizations facing today?

Lance Speck: The healthcare industry is pretty new to having data as a way to solve problems and eliminate risks as they relate to patient readmission, poor care or even improving care and treatment. The reason is that most of the different organizations only recently came into electronic records. Add to that, regulations have now forced certain standards that allow data to be communicated and systems to communicate with one another, and itís really just day one in healthcare. Other industries like financial services have been fine-tuning their analytics for the last decade. They are learning that there are hundreds of billions of dollars of savings for the customer or the patient or the member, as well as avoidance of things like fraud and waste.

There definitely are a lot of major areas that the healthcare industry can focus on. Where do you see healthcare headed as it relates to big data analytics?

Lance Speck: Weíre on day one. So there is all this data available. Everything that we see and everything that we touch is giving off signals, giving off data. The devices, everything from wearables like Fitbit to the EMR and EHRs, which are the central nervous system of a hospital, are tracking and storing data. Everybody Ė from the doctors to the patients to the payers to the back office people handling claims Ė is constantly getting this flow of data. Harnessing that data and gaining insight is the big challenge for a couple of reasons. Can they scale to be able to handle the quantity of data thatís coming through? Do they have the science? Do they know what questions to ask? Do you know what data you have, what you donít have, and whatís available to you? Can you handle the scale if you do know? And then, do you know what questions to ask of the data?

You mentioned that the financial services industry has been doing a lot with analytics over the last decade. How is the healthcare industry different with regard to data and analytics from financial services and other industries? What is unique and what is the same?

Lance Speck:
Any industry Ė I donít care if itís media or financial services or just about any industry Ė that is leveraging the data it has outperforms people in the same industry who donít leverage the data by about 5x. It does matter, and those who understand that it matters are the leaders in their respective industries. In healthcare, itís just a very immature market when it comes to leveraging data. Financial services is fine-tuning how to find fraud before or as somebody is making a purchase. And Iím not just talking about people who stole a credit card. Iím talking about third-party fraud where somebody does everything right for three years and sets up a bogus account and then commits fraud. Theyíre trying to predict where something will occur before it occurs instead of trying to chase it after it occurs. I think healthcare is far from that, so that creates a problem. After-the-fact type of fraud means you spend a lot of time chasing false positives, and when you do find a positive, trying to actually track down and recover doesnít usually have a high success rate. I think where the healthcare industry will go as it matures will be to get out in front of the occurrence of fraud. Youíve heard people talk about the cost of treatment versus the cost of prevention, and this goes for everything from fraud to a patient who has to come in and get readmitted because they werenít treated appropriately the first time. Our country, in large part, spends 80% of its money treating things that were preventable if they would have been able to predict what to do beforehand.

There is a tremendous amount of payback that you can get from big data analytics in healthcare.

Lance Speck: Hundreds of billions of dollars.

Could you give us some examples of the ways that big data analytics have impacted the healthcare industry for payers, providers and patients?

Lance Speck: On the provider side, Iíll provide a couple of examples. One will be avoiding readmission, and this isnít just a problem that people want to solve because they want to avoid letting people go and come back in worse shape. Itís also very public. You can find posted information about which hospitals have high readmission rates. Not only is it published, but there are fines that they receive from the government. So itís not just improving health, but itís also that you pay a fine if youíre not able to improve on that. Avoiding readmission can be resolved by taking data Ė every piece of data that you have available Ė and analyzing it with predictive algorithms, not just rules based on demographics. Those have been around for a long time. This is more about extending the doctor whoís looking at them and treating them personally. Itís the ability to extend what they do and what they know from their experience by giving them information that may make them ask a couple of extra questions or may make them keep the person in the hospital for an extra day. Based on the data that they have in front of them, they may also treat two patients differently even though they present themselves exactly the same. Furthermore, this is presented in a way that doctors are used to doing business. Day to day, they donít look at pivot tables and spreadsheets and reports. They want something thatís on their iPad or on their screen that says, ďHere are the top ten people you should probably worry about today, and if you expand on this, weíll tell you why.Ē They can make that call so that they can hopefully avoid having them leave the hospital and then get sick again or maybe even worse. Thatís one area.

The second one is staff optimization. Itís very difficult to know how to staff and who to staff and when so that you have the right care in place at the right time. And for anybody who has had something happen to them in a hospital or has had to go to the doctor and had to wait, itís very uncomfortable, if not dangerous. So how do you predict what staff youíll need? You can go back and look at history. Again, back to rules-based information to see what happened last year and this is how itís trending. But things are happening in the world right now as we speak that people are talking about either on social or on the news. Or the CDC can put out information about whatís heading what direction or what tragic event just happened. If people are able to apply that to their rest of the data they have Ė their historical data Ė and then run the algorithms on top of that, they can get accuracy in their staffing to have the right staff there at the right time.

Letís shift a little bit. Letís talk about the next-generation big data technologies like Hadoop. What are your thoughts on it in the healthcare space? How are successful healthcare organizations utilizing Hadoop?

Lance Speck: Usually right now healthcare technology providers Ė software and services providers Ė theyíre looking at Hadoop. Very large hospital systems like Mount Sinais, Kaisers and Mayos of the world, they have Hadoop projects going on. The large payers like Aetna, Humana and United Health all have Hadoop projects going on. Really, the shortest version of what is big data and when might you need Hadoop is usually when your data gets to a point where you canít really effectively use† it Ė when the size of your data gets unmanageable in a traditional operational database. What Hadoop does is it gives you a really solid affordable way to scale that data to any level. Across many industries, we have people who leverage the Actian Analytics Platform alongside of Hadoop because it gives them industry differentiation and uniqueness as far as scale and performance. But unlike many of the traditional software companies, it doesnít ask you to go out and throw a bunch of money by adding expensive hardware to very old software to make it work fine. Hadoop combined with the Actian Analytics Platform really gives you that scale and record-breaking performance that weíd put up against anything in the world at a reasonable cost.

On the provider side, theyíre margins have decreased every year for the past few years, so finding a dollar for them is not easy, but they have to. They have to turn it around somehow, and I think data is the key.

For healthcare organizations that are looking at Hadoop and looking at ways to get into big data analytics, where do you see the short-term opportunities for them?


Lance Speck: Find a place to start. Look at the areas of inefficiency that your group has and then try a pilot. Find an advisor that understands the space and find a pilot to advocate expansion. Itís an inexpensive way to get started, but you can see differences quickly. There was a pilot at Mt. Sinai that looked at staff optimization. They did a fairly small pilot, and they ended up reducing their overall overtime costs by 67% over a year. And yet, they saw more patients. If thatís the highest cost [staffing] and youíre able to get that out of it with relatively small investment, then thatís pretty impressive. So I would say, find something thatís causing a lot of pain thatís relatively small, and then use that to advocate for more.

When we talk about Hadoop, what Iím hearing is the hype is so great that itís solving everything. What are the myths?


Lance Speck:
The myths are that Hadoop doesnít necessarily equal big data. Thatís one thing people need to understand. Another thing they need to understand is that there are bottlenecks that you can avoid when using Hadoop. For example, we have a new next-generation way to store massive amounts of data, and yet the most common way to pull the data out today is MapReduce. That is an ice-age technique that really defeats the purpose because it slows down the performance of actually working in Hadoop. You really need to think about what it is that you want to accomplish. Storing data is one thing, but storing data without gaining insight from it is really defeating the purpose of gaining the value, gaining the efficiencies and improving healthcare. If youíre not gaining insight, youíre just hoarding. Really you want to be able to scale out. You want to be able to call that data. You want to be able to get the answers with low latency so that you can do something. And if you find out that you shouldnít have released somebody from the hospital three days after theyíre gone, thatís not as good of an answer. And if you find out that you could have had 100% greater accuracy in your data if you could have used all of the data, well thatís not good if youíre having to limit it or sample it because of what you surround Hadoop with. Or if you find out that there are 5, 6, 20 or 100 different data streams that are available to you freely that you just canít use because you canít scale or you donít have the performance to leverage it, well then thatís a shame too. So I think the myth is that Hadoop by itself will solve your problems. Rather, itís how you leverage the storage, the performance and the science to actually make a difference.

Healthcare is obviously known as a laggard in technology. Why would they leverage big data analytics now?

Lance Speck:
Well, there is the side of it where you want to make the world a better place. And then thereís the side of it where itís survival of the fittest. And, those who donít are probably not going to be around. Thatís a bold statement, but I am telling you that five years from now, those that are not leveraging the data to their advantage are going to disappear. Theyíll be gobbled up by somebody else who is, or they will just go out of business. You see it happening with the community hospitals right now. Theyíre just disappearing or being acquired because they canít deal with it. They should be able to, but itís just too hard to get the investment to actually spend any time on the data. So they fall behind more and more every day.

Is it riskier to adopt these next-generation technologies rather than stick with legacy approaches? You mentioned community hospitals. What is the cost of doing nothing? It sounds like they could be out of business.

Lance Speck: Yes. You said it. We just sat down with Childrenís Hospital of LA, and they were very adamant that doing nothing was not a safe answer. Sometimes in tech, people say itís too risky to try that new thing. But if youíre headed for a cliff, doing nothing is not good as youíre falling to the rocks. The point is that doing nothing isnít an option. This is coming. The problem is here. Costs are going up. The data is there to help you improve your care. The data is there to help you optimize your staff. The data is there to reduce readmissions and to avoid fraud. And if you do nothing with it, then youíre just asking for trouble Ė and not ten years from now, probably more like one to two.

If organizations are interested in starting their big data analytics journey, where should they begin?

Lance Speck: I think first they should find an advisor, maybe somebody who has had success in the past to give them a ďwhere to startĒ blueprint or roadmap. Then I think that person would say, ďLetís not solve all the problems that have been around for decades. Letís just find one and try to focus on something small first. And then, as I said earlier, use that to advocate for bigger changes in the future. I really think that they should take that first step. I always challenge people when Iím done speaking or meeting with them to go back to their organization and say, ďAre we applying any science to the data that we have?Ē If they say yes, then theyíve taken the first step. The answer, though, is probably going to be no. The second question they should ask is ďWhere do you think we feel the most pain as an organization or our members or patients feel the most pain? ďAnd the answer is usually ďsomewhere around x.Ē The third question is ďDo you think there is some way that weíd be able to predict the way to avoid that using data?Ē Those three questions alone have started the journey. The journey is the key part of your question. Donít think there is a magic button or magic pill or silver bullet. You have to see this as a journey that youíre going to continue for as long as youíre in the space, and you just need to continue to improve on it by gaining access to more data, by adding better predictive algorithms, and continuing to improve over time.

Thank you, Lance, for providing this great insight into how big data and analytics are changing the healthcare industry today.

SOURCE: Big Data Analytics for Healthcare: A Q&A with Lance Speck of Actian

  • Ron PowellRon Powell
    Ron is an independent analyst, consultant and editorial expert with extensive knowledge and experience in business intelligence, big data, analytics and data warehousing. Currently president of Powell Interactive Media, which specializes in consulting and podcast services, he is also Executive Producer of The World Transformed Fast Forward series. In 2004, Ron founded the BeyeNETWORK, which was acquired by Tech Target in 2010.† Prior to the founding of the BeyeNETWORK, Ron was cofounder, publisher and editorial director of DM Review (now Information Management). He maintains an expert channel and blog on the BeyeNETWORK and may be contacted by email at†rpowell@powellinteractivemedia.com.

    More articles and Ron's blog can be found in his BeyeNETWORK expert channel.

Recent articles by Ron Powell

 

Comments

Want to post a comment? Login or become a member today!

Be the first to comment!