How to Reduce Healthcare Fraud with AI

Aleksandar Lazarevic, Senior Director of Data Science at Aetna

How to Reduce Healthcare Fraud with AI

Utilizing analytics to reduce fraud in healthcare

How to Reduce Healthcare Fraud with AI

Thanks to Our Podcast Sponsor, Noveta Health

Save 25-30% on Your Health Benefit Spend

Recommended Book:


Best Way to Contact Aleks:


Mentioned Link:

Personal Website

How to Reduce Healthcare Fraud with AI with Aleksandar Lazarevic, Senior Director of Data Science at Aetna (transcribed by Sonix)

Hey Saul Marquez: Outcomes Rocket listeners thanks for tuning into the podcast again. Tired of your businesses healthcare costs unpredictably increasing every year? Healthcare costs are typically a business’s second or third line item expense. And if you’re like most employers, it’s an expense that’s growing faster than your revenue. Luckily for employers Noveta Health has the solution. Noveta Health is a full service healthcare consulting firm with proven strategies to lower your health care costs by up to 30% or more. They operate on a fee for service model and never mark up any of their medical or pharmaceutical claims. None of your employees have to leave their doctor or pharmacist either. Their health captive and pharmacy benefit manager are the most cost effective and transparent solutions in the whole country. What they do is not magic. It’s just honest. So if you’re tired of overspending on health insurance and want to learn more visit outcomesrocket.health/save for a free spend analysis to see how you too could save by switching to Noveta Health. That’s outcomesrocket.health/save for your free spend analysis outcomesrocket.health/save.

Saul Marquez: And welcome back to the podcast. Truly appreciate you tuning in again. Today I have a very special guest. His name is Aleks Lazarevic. he is the Senior Director of Data Science at Aetna. Aleksandar is responsible for overall predictive analytics solution in Healthcare Fraud Initiative at Aetna. In addition to health care industry experience, he has extensive experience in various data analytic projects ranging from banking credit and insurance industry to diagnostics and computer security applications. He’s co-editor of a book on cyber security threats written eight book chapters and published over 50 research articles which recited more than three thousand times. He holds a P H D degree and data mining machine learning from Temple University and he has been frequently giving presentations on these topics of predictive analytics data mining conferences in the past. So it’s a true pleasure to have Aleks on the podcast. Given that today we’re really spending a lot of time in machine learning and predictive analytics for healthcare. So I’m excited to dive in to his expertise and some of his thoughts on these topics and healthcare so Aleks welcome.

Aleksandar Lazarevic: Thank you for having me Saul. It’s really my pleasure to be here and talk to you on the Outcomes Rocket podcast I think this actually was great. Thank you very much.

Saul Marquez: All righty. No that’s great Aleks. So what is it that got you into health care. I mean you could have landed in a lot of places and you dabbled in some but what is it that ultimately got you interested in health care?

Aleksandar Lazarevic: So actually my passion as a kid was always math and computers. However my mom always wanted me to be a doctor so then I came to the U.S. to pursue my PH degree. I started to study machine learning data science and after spending several years across several industries as you mentioned I settle healthcare which was a perfect fit between what my mom wanted and actually what my aspirations actually joking aside I chose actually health care because I honestly believe that there are so many opportunities how big data I really think something that I’m good at could help our broken health care system decisions in the health care industry today better are doctor decisions so that this is decisions made with health care payer like health insurance companies are mostly coming from the expert knowledge and I believe the data has not been used so much historically but it can discover important factors and it can improve the oral health. That’s basically why I’m in healthcare.

Saul Marquez: I love it love it. Yeah it’s a great tool. I feel like we’re learning we’re learning more and leveraging it more. What would you say for the health leaders listening to this podcast today. What is that hot topic that needs to be on their mind as it relates to data and machine learning and how are you approaching it.

Aleksandar Lazarevic: So I believe the May the hot topic for the leaders today is actually our overall healthcare system. If you look this historically in 96 this they probably spend 2 to 4 percent of GDP on the healthcare and average healthcare spending per person was one hundred fifty dollars. Today we are spendin, the United States is spending more than three point five trillion dollars on an annual basis on health care which is 18 percent of GDP. And this actually this is enormous increase over just period of 40 to 50 years. U.S. extra spends four times more money or healthcare than any other country in the world. And at the same time according to the Institute of Medicine one third of this money is based on our population is becoming older and sicker. Sixty percent of population has at least one chronic conditions and almost 90 percent of the course actually is coming from treating those individuals with chronic conditions. So despite all this spending the US is still rack the last thing the health care quality almost 34 developed countries and this is simply unsustainable in my opinion. Our health care system is simply reaching a breakpoint. These are actually the fact that they believe that every medical leader should have in their mind an extra they should think how we can do and what we can do to fix this broken health care system. So I believe it’s one of the major problems in our health care system is that because defensive medicine and at the same time reactive medicine. I believe even Benjamin Franklin mentioned that ounce of prevention is worth like a pound of cure. And we need definitely to move towards preventive health care and adopt more holistic view of what actually health is. And this is actually by according to World Health Organization health is defined as a state of complete physical mental and social well-being and not just the absence of disease or infirmity. So if we adopt this definition actually health care actually becomes what happens between doctors visit and not merely what is in the doctor’s office or in the clinic. So I think it’s far more important. Where do you live and how do you live your life then actually what kind of healthcare is provided to you. So that’s I believe that’s where we need to go as a health care society. And I believe that’s basically what we need to do. So in my opinion big data that you picks will play a significant role there because I honestly believe that using the data using data driven evidence based approach would definitely help our healthcare system.

Saul Marquez: Yeah. Now very insightful Aleks and and yes some of the stats you’ve mentioned three point five trillion, third of the money is wasted,last data thirty four, defensive and I love your quote overall like you know an ounce of prevention is worth a pound of care. That definitely resonates and it’s true right. I mean we’re definitely very defensive in how we do things. We’d love to hear a story from you of how you guys are working to use whatever is thoughtful approach in making better results for populations.

Aleksandar Lazarevic: So Aetna and actually right now Aetna with CBS is actually to try to follow this main idea of preventative health care and focus heavily on our members and how to best engage with them. Healthcare is becoming increasingly customer focused. And basically you’re trying to leave providers and appears with the challenge and leverage these big data to personalize care. So basically what we need to do is not only to do preventative care but also to personalize the care to our members because that’s what it is. We cannot assume that we have healthcare which is safe for everyone. The needs of people are different and the emergence of these big tech players like Amazon, JP Morgan, Google Microsoft in the healthcare space it doesn’t make a lot of evidence that they’re actually customer needs that they’re not met. And there is a big opportunity there. Customer in healthcare face for neglected for so long. And I believe this is finally time to change that. The merger between CBS and that actually allows us to actually to try to put two members on a central place and we can try to basically help them live healthy life. That’s what we’re trying to do. So since I mentioned that I would like to actually mention two main initiatives that Aetna analytics organization has. Aetna as far the actually centralized organization we are providing support for Aetna. And our two actually main initiatives actually as I mentioned how to better engage with our members, improved our health and satisfaction but also how to improve the provider quality of care by providing personalized care through precision medicine also improving provider efficiency and also reducing the health care fraud based reviews. We are trying to do this by moving to a value based care. We’re joined by interests because of its elected health care providers. You know if you’re okay I can talk a little bit more about actually how we are addressing the health care fraud basically abuse as one example. What we’re trying to do.

Saul Marquez: Sure, let’s do.

Aleksandar Lazarevic: Okay. So we actually started the project maybe like few years ago. Before that Aetna was using a vendor to address health care fraud and abuse and they were not happy with the performers. They were not happy with the price of course and we actually jumped on this opportunity because we believe that big data like this could really help us quickly and accurately identify those fraudulent providers. I mentioned earlier that the U.S. spends like 2.5 billion dollars every year and nobody actually knows how much of this money is embezzled. You’ll do fraud and abuse but according to some statistics from FBI and some industry payers between 3 and 7 percent of overall spending is actually lost to fraud based on abuse. And this is significant amount of money if you convert to. This is between one kind of building and 350 billion per year. This is a huge money and actually you’re losing a lot of money because of that. And as you know fraud is happening every day I’m receiving emails on a daily basis about different fraud schemes and I’m just a year ago we had an example of that of a protease that was more than 1 billion dollars so it’s not uncommon anymore to see that actually this large fraud cases are happening and it’s actually even worse if you consider the crisis. That’s because our national opioid crisis and many players see this as an opportunity to make quick money. So many people perceive health care fraud as a lucrative business. It’s so lucrative that actually even some criminal drug dealers are switching to dealing with healthcare fraud because money is much better, work is safer and the penalties are much lighter. You will not be killed for doing health care fraud right. Very often actually Medicare and Medicaid detectives and investigators active in their making arrests that they are finding those profiles of many are asking this kind of drug dealers. This is really really amazing what’s happening and how much money we are losing.

Saul Marquez: Aleks quick question for you then. So with the types of fraud I mean what exactly is happening? Our people. What kind of fraud is it? Is Medicaid fraud? Is it Medicare fraud? Is it insurance fraud? All of the above.

Aleksandar Lazarevic: All of the above.

Saul Marquez: OK. OK. Got it. Got it.

Aleksandar Lazarevic: So basically fraud…

Saul Marquez: However they can get it?

Aleksandar Lazarevic: Fraud actually can range from many different initiatives and actually fraud is actually defined as intentional deception of the system which is basically in order to prove that some healthcare providers or doctors are doing fraud because to prove that intent. And that’s not easy to do because you have to look historically what these providers are trying to do and trying to establish the petrol through data and then to prove something is fraud. Waste and abuse is somewhere in the middle so between from fraud to like a normal behavior. So basically waste and abuse could be counted as improper actions trying to bend the rules, think some inefficiency or just simply unintentional or incorrect payments on the tapes. So all of this kind of belongs to overall base that we have in healthcare fraud. So then we started the project actually our initial focus was mostly fraud and actually at that time we had a big Hadoop infrastructure. We were trying to collect all the data mostly around medical claims how we pay our claims what the data doctors what kind of procedures they are doing. We actually had a business partner which was a special investigation unit. These are people who are actually doing investigation or full perpetual leads as well as suspicious healthcare providers and actually as of one of the first steps we actually created a suite of litigation tools that can help them doing that and investigations much faster and more efficient.

Saul Marquez: That’s great.

Aleksandar Lazarevic: So on top of that we actually had to run some analytics of course because the whole hope of action applying big data analytics in healthcare fraud detection and how we approach this, we actually had three major Brooks teams we try to actually first drives the low hanging fruit some fraud schemes that say you for a special investigation unit investigators already familiar with and we tied to this kind of quick coding of these rules. We call those business rules and these rules actually are defining the exact behavior that they’re familiar with however very often providers when they figure out they are detected on they are kind of suspicious that they could change themselves and for this kind of behavior we are trying to apply or actually use some kind of machine learning to take these kind of deviations of these kind of known behavior.

Saul Marquez: Yeah.

Aleksandar Lazarevic: And very often you don’t know about some schemes because these close games are developing pretty quickly and in order to address some of these schemes to the outside they actually tried some of anomaly detection techniques in which we are trying to detect all deviations from the normal or norm because that’s how we try to actually structure although we think that people are running for the protection.

Saul Marquez: Interesting.

Aleksandar Lazarevic: And so far the actual results were pretty good, in the first year we were able to save around a million dollars. In the second year it was like 40 and in the second finally at 30 I see it was 82 million or sexually were pretty able to almost double this every year.

Saul Marquez: Man.

Aleksandar Lazarevic: And we’re not sure how much we will be able to continue that. But we’ll see.

Saul Marquez: That’s amazing. I mean those are some serious numbers.

Aleksandar Lazarevic: Yes these are serious numbers. And of course we still believe there is much more to detect. And right now we are actually planning to expand more into the based on abuse games which actually we don’t need to prove exact intent from the providers and also to put more focus on the prepay than actually postpaid. What typically happens in the fraud investigation you have to actually analyze historical behavior provider, you can request some medical records in order to show that actually whatever they submit for claims and whatever they tell the medical record does not match and actually then only you can actually flag that provider for fraud and start denying those claims in order to actually to minimize that investigation time you’re trying to do everything in a prepaid and you can do a quick review and everything is done in prepaid instead of just waiting for six months to do all these things because we’ll be basically electrocuted. So that is basically to trying to do and we believe that could further actually increase our safety.

Saul Marquez: Fascinating. Well it’s a great thing that you guys are working on this project Aleks. And as the technology is there to leverage, to make things better, I think it’s important to realize listeners that hey you know that it’s being leveraged for bad reasons too. And so if you don’t stay ahead of it you’ve got you’re going to be behind. So Aleks has given us some great examples of how they’ve been able to effectively implement some of these machine learning techniques. Data analytic techniques to save eighty two million dollars a year. And that’s tremendous work by you and your team Aleks. Give us an example of a setback you guys had and what you learned from it that has made you better.

Aleksandar Lazarevic: Ok. So I can go back again to the road project when we started that project. We actually had lots of data. We started with the pure data science field. And at that time and actually still today data science was called field as you know data scientists were the sexiest job of 21st century…

Saul Marquez: For sure.

Aleksandar Lazarevic: And because we were actually really hungry to apply some of this advice we’re still learning or data science to actually sold or identified those broadband providers. Pretty quickly we actually realized that the outcomes that we were actually providing to our business partner were actually mostly wrong and not well received. So what we forgot in all that hype and excitement of doing data science is actually about our business partners and actually their needs.

Saul Marquez: Yeah.

Aleksandar Lazarevic: It simply didn’t occur to us that actually you investigators are no data scientist and you don’t speak the same language. So we were actually talking about prediction performance of our models, the Postal Authority, standard deviations. And they simply were talking about for example what TPP procedure the doctors were doing, what kind of overall it’s a limited amount they provided to us. What did the J Cause and all this kind of stuff. We actually started to pay closer attention to what they want and what they need. We were trying to sit with them and understand the thought process behind the investigation and we try actually to incorporate that process is all about analytics and that’s how we came out with that with those business rules to capture the loan fraud schemes first and then try to expand it. That’s how we structure our law ethics around that kind of learning. Actually our failure initial failed. So there’s a major lesson for us that well is actually that you always need to listen what the business needs from your business partner. You’ll just need to keep them in the loop during the development of your overall analytical process and solution. And finally you need to keep them in the loop when you finally evaluate the results together because you need to align with all of this what you do not need is actually you don’t need a solid to listen to what they tell you what you need to do because since they are not data sciences they’re not the medical people, they may not have a good kind of knew what is possible from the analytical point of view. So basically I believe that the role of the data scientist or people scientist unicorn would be actually to understand what the business problem is to figure out what is the right data sources to chase to figure out what is the right analytical solution and also call to deploy that solution. And that’s basic I believe the biggest role of data scientist is not just a technical person it’s much more than that. And that’s why many people are reporting to data science unicorns.

Saul Marquez: Love it. Now I think it’s a great great example and a great story. Yeah it gets exciting right. You get technologies like these and then when the rubber meets the road, theory needs to mesh with the frontline. And so…

Aleksandar Lazarevic: Yeah true true, very true.

Saul Marquez: You know and I’ve had the privilege of talking to a lot of really smart folks like yourself Aleks that are just brilliant with these technologies, data scientists and yeah it’s that that meeting of the theory with the rubber at the road and it sounds like you guys did a really great job of recovering and enhance the success of the project.

Aleksandar Lazarevic: Yeah yeah you’re absolutely right there.

Saul Marquez: What would you say one of your proudest leadership experiences has been to date?

Aleksandar Lazarevic: I think I’ve been remodeling the fraud waste and abuse section, we started as I mentioned during your protest action but there’s a natural extension of this work. We actually started to worry about provider efficiency, provider quality of care, and third based in jello. So in addition to actually indexing waste and abuse and all this kind of bending the rules, incorrect payments, we actually started to look what is the whole journey that our members have been they go through the treatment for particular health problems for example cancer, obese the or something else. So we were actually trying to collect all the claims that they had during that process. We tried to understand what is at treatment that they’re going through and whether healthcare providers are using the most efficient treatments available or they are trying to do some kind of unnecessary service. So we actually developed was we did cool way of visualizing those people but we were able to actually immediately notice all possible deviations from the standard treatment. And in addition when we incorporated the members health conditions we were actually able to see actually how we can provide the combination to the doctors based on this kind of data which I call like a data driven recommendations. What is that optimal or personalized healthcare. What we can do because not every person will need the same care. Right. So basically…

Saul Marquez: Right.

Aleksandar Lazarevic: By doing this analysis by looking all these people who went through this process and if you’re a health insurance company you have a mini or so claims and minutes of people and by doing this relative to what we really have the capability to look what works what person. And based on this combination will be able to recommend this to the doctor. There is no need for doctor anymore to beat for journal article to see what kind of techniques are there. We can all be in that position. We can try to look at the data. We can try to see what’s working and what’s not working. And Baker recommend this to our members. So this is something that I was really proud that we were able to develop and we actually tried to address this kind of providing optimal personalized care to all people, try actually the same time be more cost effective and in the same primary we actually be able to actually to optimize the network our networks by actually eliminating those inefficient healthcare providers but also we were able to address through this historical analysis if we would put those very important problem in healthcare which is cost fallacy because very often people do not know how much money they would pay. If you take into consideration the member characteristics if you know the documentation for the doctor, it would be pretty much able to figure out how much the members who pay at the end of the trip something that does not exist today.

Saul Marquez: And then these analytics are shared with the member when they’re making a decision of who to go to?

Aleksandar Lazarevic: Yes they could be shared with a member they can make decisions based on the quality of care provided by the doctor, they can make decisions based on the course, they’re going to pay at the end and they can make decision whether these providers a network on top.

Saul Marquez: Love it super useful.

Aleksandar Lazarevic: Yes yes. I think this is based. I’m really proud that my team was able to kind of do this kind of work and I believe this is really great opportunity there to extend this work as well.

Saul Marquez: And it gets back to what you said at the beginning of our interview Aleks about the consumerism coming into healthcare right. I mean with large deductibles and just more and more of the financial burden being shouldered by the individual you’re going to have to start doing these types of things.

Aleksandar Lazarevic: Absolutely. Absolutely. And actually there will be more and more especially for a new generation of millennials. They are more kind of celibate for the money and how much they pay. And this will be more and more important things that need to take care and we need to consider.

Saul Marquez: Love it man. Great example definitely one to be proud of. What about an exciting project that you’re focused on today?

Aleksandar Lazarevic: So I think the one that I just mentioned is definitely very exciting at the moment. And I think we are proud of that. And as I mentioned I think this is probably the best way to improve the healthcare outcome in some way actually to improve the overall health, better engage with our own members because this is uniquely to show actually our members that they are not just here to pay their bills. They actually here through the whole journey with that because we could be there when recommending what doctors think, we could be there and actually recommending what treatments would be because we could provide the same kind of recommendations to them and to doctors so they can have informed discussions with the doctor and also we can help them actually with how much money that whole treatment would be. So I think we’d better engage with our members as something that we are really trying to do to improve that kind of overall satisfaction of our members. And the final goal we want to actually have our members healthier because at the end we will make more money by doing that right.

Saul Marquez: Absolutely. Absolutely. Now this is really good stuff. Aleks appreciate you sharing that. getting close to the end of the interview here, let’s pretend you and I are building a leadership course on what it takes to be successful in healthcare. The ABC’s of Aleks and so it’s a mini syllabus, lightning round style and ask you five questions followed by a book that you recommend to the listeners. You ready?

Aleksandar Lazarevic: Yep. Okay.

Saul Marquez: All right. What is the best way to improve healthcare outcomes?

Aleksandar Lazarevic: I reiterate myself again. I think best way to improve outcomes is to use data driven or evidence based medicine to leverage all the data that we have available. I know in healthcare that’s not quite possible yet but if we are able to share all these data from all these different hospitals from all these healthcare resources, we will be able to have enormous knowledge of what’s going on what’s working what’s what’s working and what’s basically what’s efficient. I think data is the basically new actually oil for the healthcare industry.

Saul Marquez: What’s the biggest mistake or pitfall to avoid?

Aleksandar Lazarevic: I think biggest mistake would be to work alone and come to a quick conclusion without consulting other healthcare players or your colleagues or business partners.

Saul Marquez: Love it. How do you stay relevant despite constant change?

Aleksandar Lazarevic: Learning learning learning. I think that’s the issue as the only way because both fields both healthcare and big data analytics are changing so fast that you would definitely have to read a lot to see what’s going on. You’re probably aware that so many healthcare startups are doing different things and trying to address the people healthcare. So basically by speed by trying to learn more you will definitely stay relevant.

Saul Marquez: What’s one area of focus that drives everything in your organization?

Aleksandar Lazarevic: I believe member engagement. How to put members in a central place and tell them basically live healthier life.

Saul Marquez: So the last question that I have here is a two part question. What is your number one health habit and what is your number one success habit?

Aleksandar Lazarevic: Number one health habit is probably to have 10,000 steps everyday.

Saul Marquez: Oh nice.

Aleksandar Lazarevic: Although I’m not very physically active except I’m running up to my kids that the earliest. So that’s pretty good. Quite often.

Saul Marquez: It’s really good.

Aleksandar Lazarevic: I’m usually successfully doing this. I think if you if your body is active you don’t need to do actually all these kind of physical exercise and everything. It’s just important not to be like sleeping all the day.

Saul Marquez: Sedentary.

Aleksandar Lazarevic: Yeah yeah exactly. So that’s basically you’re getting my health habit. And of course I can… three things that I usually tell people, sleep well, eat healthy, and try not to be exposed to such stress too much.

Saul Marquez: Love it. Now that’s a great, great ones.

Aleksandar Lazarevic: And regarding the…

Saul Marquez: Yes success habit.

Aleksandar Lazarevic: Success habit right to dedicate some time for yourself because…

Saul Marquez: It’s good one.

Aleksandar Lazarevic: You’re going to have all this time to meet the thought of people but you rarely have time to be with yourself. And actually if you want to really do something great you really need to focus on something and I expect when you’re alone. Bedspread what I’m always trying to do to date a few hours of my day just to do something for myself.

Saul Marquez: Love that. Aleks I love that. I believe it and practice it. It’s so so key, folks you could find this mini syllabus as well as a full transcript, links to the things that we’ve discussed go to outcomesrocket.health and in the search bar. Type in Aleks Lazarevic or type in aetna and you’ll find this episode will come up. What book would you recommend to the listeners?

Aleksandar Lazarevic: When I was younger actually my favorite author was Hermann Hesse and one of my favorite books from him was Siddhartha. It’s actually a spiritual journey of a young man during the time of when Buddhism appeared recently I don’t have time to read that much as I would like but that absolutely must for all people. However I still spend some time with my kids reading some of the tiller books and one of my favorite they are is actually the Lorax by Dr. Seuss. I think even adults can learn a lot from that book.

Saul Marquez: I love it. I love it. Lorax. That’s great. Lorax and Siddhartha folks get those on your reading list. Aleks before we conclude I’d love if you could just share a closing thought and then the best place for the listeners could follow you or get in touch.

Aleksandar Lazarevic: So Okay the best place to get in touch with me is definitely in Linkedin. I still keep my personal website and my old university University of South SNC if you just type my name on Google you will find me. Final thought is that Jill Carroll is finally ready for disruption and I believe that big data analytics, machine learning, or some people would even say A.I. would play a significant role there. So that would be my final thought.

Saul Marquez: Love it Aleks. Hey keep up the great work. I really really appreciate you sharing your thoughts here today and definitely looking forward to staying in touch. Thanks again.

Aleksandar Lazarevic: Thank you so I really enjoy our conversation. Thank you very much.

Thanks for listening to the Outcomes Rocket podcast. Be sure to visit us on the web at www.outcomesrocket.com for the show notes, resources, inspiration, and so much more.

Sonix is the best online audio transcription software in 2019.

The above audio transcript of “How to Reduce Healthcare Fraud with AI with Aleksandar Lazarevic, Senior Director of Data Science at Aetna ” was transcribed by the best audio transcription service called Sonix. If you have to convert audio to text in 2019, then you should try Sonix. Transcribing audio files is painful. Sonix makes it fast, easy, and affordable. I love using Sonix to transcribe my audio files.