Corporate Innovator: Tina Moen, Deputy Chief Health Officer at IBM Watson Health

Steven Loeb · October 22, 2018 · Short URL:

IBM's initiatives include Care Manager and Watson for Genomics to individualize care

While most entrepreneurs want to be the one to discover the next Amazon or Twitter, oftentimes major technological shifts are coming from the big companies, the players that have been on the scene for years, if not decades. Those companies have survived because they know how to pivot. They're the ones who either seed new ideas or acquire them and distribute them. 

In this column, we talk to those companies and their innovators who are preparing them for what's coming.

Healthcare is one areas where advances in technologies like big data, artificial intelligence and machine learning are having a major impact on how the space operates.

IBM is a company that has been making big strides in some of these areas, with products and service such as Care Manager, which individualizes patient care needs; Watson for Genomics, which uses AI for precision oncology; and Social Program Management, which supports government agencies in their work to deliver health and human services.

I spoke to Tina Moen, Deputy Chief Health Officer at IBM Watson Health, about some of the ways that AI is changing healthcare, some of IBMs initiatives toward applying technology to the space and what's next in that evolution.

VatorNews: Tell me a little bit about yourself and your role at IBM.

Tina Moen: I am one of the Deputy Chief Health Officers here at Watson Health, PharmaD by background. I spent the early part of my career as a clinical practice patient in a variety of patient populations, starting in pediatrics and ending up in hospice and HIV and transplant. So, I really saw the spectrum of the kinds of patients that often need a lot of support and, even way back then, a bazillion years ago, the amount of data and information and content and patient demographics and all of that stuff was available, but in 50,000 places. That realization was often a frustration for me as a practicing clinician and for many of my colleagues as well, so when the opportunity arose to move my career to healthcare technology, I took the opportunity, a little but nervously but excitedly, and have been in healthtech for now 16 years or so.

I really found my fit. I’m in this space because, as a clinician, with the caregiver nature that clinicians have, I am really excited by the fact that, in a role like this, and a company like this, my ability to impact the health of people, not only in the U.S. but all around the world is much greater than it would have been in individual practice. While certainly at more of a distance, so you sometimes miss that, the ability for me to impact patient care around the world is what keeps me in healthtech. It’s pretty exciting.

VN: How has AI in healthcare evolved? What couldn't we do five years ago that we can do now from the perspective of caregivers/patients?

TM: It’s such an interesting question because I think in every industry there’s evolution, that’s just the nature of life and healthcare is no different. But the interesting thing is that healthcare is, as I mentioned, at its core, a patient-centered, a human-centered industry. So, oftentimes I have these interesting conversations with people where they’re like, ‘You say it’s a human-centered industry but yet you work in technology,’ and I actually often say, ‘That’s the perfect combination. Because healthcare is a human-centered industry, and we’re trying to help people who are taking care of patients, and help them make decisions, and help them make the best recommendations to those patients, which is all of us. The ability for technology to help that decision maker make the best decision for that patient, help that patient make the best decision for themselves, that’s exactly where technology should be sitting.’ That’s the evolution I’ve been watching over the last five years. As the healthcare industry has made really amazing discoveries and advancements in how we take care of patients and different procedures and surgeries and medication options and all sorts of things, that just grows and grows and grows the data and the content and the insight that’s available. That makes it so that every time something new is developed or discovered, that’s one more thing a human being doesn’t have to keep in their mind because it just is not possible; we all can’t possibly remember all the things we’re asked to remember on a daily basis in our lives in general, and then you put somebody in a clinical field, or a healthcare field, even if it’s on the administrative side, and the amount of data they’re expected to make decisions with is just exploding.

So, we apply analytics to the data and then AI, as the evolution of analytics really is what we’ve seen evolve. It’s helping to transfer that immense amount of data and the knowledge that’s out there and available to help people make decisions into a consumable, accessible format that really leads to insights, so those people can make the right decision in whatever decisions capacity they’re in. Let’s applaud all of the people who’ve made these great discoveries and then let’s apply technology to help those discoveries get to that point of care, get to that decision maker in the administrative office, get to that decision maker in a life sciences organization who needs to decision what the next medication might be.

VN: Can you provide specific case studies where AI was applied successfully, meaning better health outcomes?

TM: I’m not sure how much you know about the clinical trial process, but clinical trials are complex algorithms to include or exclude someone, whether they’re appropriate for a trial or not. Really complicated algorithms and, oftentimes, it’s very hard to fill clinical trials with eligible candidates where that medication might be the thing that they need. They may be running out of options, so how can we use AI to sift through that massive amount of data of what’s needed to be eligible for this criteria, what’s reality for that patient, what personal characteristics do they have, and where do those things match up? Do the things that have to match up actually match up? Are there some variable elements? Let’s highlight those and say, ‘If you lower your blood pressure or lose some weight you’ll be eligible for this trial.' How do we use technology to match that data up and make recommendations of who’s appropriate for a trial at a speed and a scale that human beings just haven’t been able to do? We had a really interesting partnership with the Mayo Clinic on this and have seen that our AI technologies have helped increase their enrollment in breast cancer clinical trials by 80 percent, so a humongous number. All of those people may not otherwise have been identified as appropriate. This is a perfect example of how AI helped the decision makers, the clinicians who are making recommendations for those patients, make sense of the data, capitalize on all of the information that’s out there and identify patients who are most appropriate for these trials. It's really fascinating.

Another example that I always love to talk about is our work with Medtronic around a solution called Sugar IQ. Diabetes, as we know, there’s a humongous patient population that struggles with that every day. Using AI and analytics and the data from this app, we’ve helped patients stay in an appropriate glucose range for 36 more minutes. 36 minutes may not sound like much unless you’re diabetic, and that identifies a positive trend, and then we can make it 37 tomorrow, we can make it 38 the day after. How do I, as a patient, interact with this cognitive app that’s making sense of the data that has access to? My glucose levels, how much insulin I’ve been given through their glucose monitor, the food that I’m eating, the exercise that I’m doing. Again, making sense of all of that data in real-time and sending that feedback to our patients in their real lives, when they’re out there in the real world, and saying, ‘These decisions lead to staying in a healthy glucose range and these decisions don’t.’ So, it’s really fascinating the myriad ways we can apply AI to healthcare.

VN: I’ve heard some people say that machines can’t replace doctors because doctors have to be warm. A machine can’t sit with a patient and comfort them, for example, but AI could be used for something like diagnosis, for example. Is that how you see AI?

TM: Certainly our approach to AI and healthcare, we use the phrase, ‘augmented intelligence.’ So when we talk about AI that’s what we’re talking about, with a couple of tweaks to what you said. A big one is that our goal, our focus, is to provide a set of evidence-based recommendations to a decision make, so not to make the diagnosis. We don’t create solutions that are intended to diagnose, for a variety of reasons. There is no substitution for standing in front of a patient and looking at them when you’re making these final decisions. There’s numbers and data that make it look like it’s one thing, but you look and you say, ‘Yeah, but there’s something else going on.’ So, we do not diagnose, we make recommendations and it’s a human being, a decision maker, who uses that to augment their intelligence and help them make the best decision for that patient that they are interacting with in a live manner.

The other thing to think about is this idea of decision maker. Oftentimes it’s a physician, of course, but there’s lots of decisions that are made in the healthcare industry by clinicians and nonclinical folks, whether it’s on the population health management side, the administrative side of a provider organization. This idea of all the decisions that have to be made along the continuum of the healthcare journey, if you will, so start way at the beginning of drug discovery and life sciences and go all the way through everything that happens with payers and with governments and the provider, all of those segments, the decision makers all along the way all face the same data challenges. From different angles, of course, and trying to answer slightly different questions, but that explosion of data and that idea that putting multiple data sets together and driving insights out of it, that’s what AI can help with. It can help decision makers make sense of the data they have access to so that it will augment their ability to make the most effective decision.

VN: Talk to me about IBM Watson Care Manager, which helps to individualize patient care needs. How does it work?

TM: It’s fascinating and it’s one of my favorite of the things we do because it does a really good job of supporting the acknowledgment that a person’s health is much broader than the time they’re in a doctor’s office or in hospital. And, as a clinician, I would much rather be able to help a patient, be it myself, my family, my loved ones, your family, your loved ones, be effectively healthy outside of a doctor’s office and outside of a hospital. We don’t want it to get to that place. And, often times, when people think of healthcare they think about a doctor’s office or a hospital. I want to note that we’re 'Watson Health' within IBM, as opposed to 'Watson Healthcare,' because we’re looking at the health of individuals as opposed to just the healthcare component of their lives. Care Manager helps support that.

If you think about things like social determinants of health, and all of the factors that go into keeping someone healthy, 70 percent of them are not what we think of as traditional healthcare. So, 70 percent of what goes into making us healthier are things like housing and food and behavioral health and things like that. Care Manager has the same sort of foundational capability where all of these data sets that potentially represent a person’s health. What zip code do they live in? What education opportunities do they have? Are they in a food desert? All of these things and data sets are brought together and can be interrogated in the context of each other. So, someone may be a diabetic, for example, in an area that might have access to healthy foods and maybe the person has a very high education level. That same diabetes in someone who is the flip of all of those, those two people have much different risks, ability to maintain a healthy glucose level and effectively maintain their diabetes. Care Manager helps those people who look at that kind of stuff.

A really interesting application we’ve used with it is in the court system, actually. If you think of a court system you probably don’t think of healthcare but we have this great partnership with Judge Anthony Capizzi. He’s a juvenile court judge and he, like a clinician in a clinical setting, has less time to spend with each child than he would probably like, but what an important role he plays helping this child navigate his or her way through the court system. He has an opportunity to say, ‘How can I help you most effectively maintain or improve your life condition?’ part of which, potentially, could be health. So, Care Manager looks at a vast amount of data related to this person’s case, or situation, through 30 to 300 pages of paperwork per child that judges have to review prior to seeing this person. Our solution does that for them in very short order, of course, and creates a dashboard of the most important information. How is this kid’s housing situation? How is this kid’s education opportunities? And things that would go into those social determinants of health also go into how he can help that child effectively move through and, eventually, moving out of the court system in a whole in a healthy and safe manner.

If you use that court example, which I think is so fascinating, and apply that to all of the areas of our lives that ultimately impact our health, which, as a clinician, is all of them, your work impacts your health, your social life impacts your health, your health is a continuum through all of it. Care Manager can pull those various data sets in and then make sense of it in context and product insights from disparate data for a decision maker.

VN: Can you tell me any of the results of Care Manager being used in the court setting? What kind of outcomes have you seen?

TM: Much like a physician, Judge Capizzi is seeing 20 or 40 kids a day, whatever his docket might look like, and he’s got 30 minutes with each of them, much like a physician with his patient load. The idea that there’s 30 to 300 pages of data that he may need to review before each of those children, it’s just impossible. So, he has talked about how the dashboard helps him spend five minutes preparing to talk with the kid, and being able to save the rest of that time to actually have that human connection and use the data to help him make decisions, as opposed to using all his time to review the data.

There’s such a strong correlation between all ages of folks in court systems and, if you think of health broadly as a variety of health elements, including mental health, behavioral health, but then even, as I mentioned, some of the social determinants of what’s their housing situation look like or do they access to healthy foods, and things like that. It’s a really interesting case study.

VN: You have IBM Watson for Genomics. Where do you see valuable new insights being applied to actionable items?

TM: Again, think about the foundational level being this vast amount of data. Certainly genomics is a growing body of data, so there’s a couple of elements to that. Not only is it a large data set for decision makers to have to consider and keep in mind, but it’s also constantly changing. So, what if my patient today would benefit from something that was just published last week that’s sitting on my desk at home and I haven’t gotten a chance to review it and add it to my arsenal of knowledge yet? Genomics is a perfect example of not just being able to recall information that is needed but also keeping up with newly published things.

If you think about AI laying across the top of large data sets, diverse data sets, that are coming from different places, sometimes they’re structured data and sometimes they’re unstructured data. The vast majority of healthcare, 80 percent, is unstructured, and often times that has been difficult to incorporate into technology solutions without the use of NLP and things like that, which can make sense of that data in a computerized way. Take those vast data stores, and disparate data stores, and apply technology to it to answer specific questions; in the case of genomics it’s looking at what their sequencing is and bumping that up against this body of data, and saying, ‘Here’s what your sequence data looks like, how does that connect to the genomic and scientific data that we have? The pharmacological insights that we have?’ We can uncover a variety of options that could potentially be appropriate, and, of course, a variety that are inappropriate, and being able to sift through all that and highlight options so that the decision maker, again, can use their time with the patient, particularly when that is being applied to cancer patients. That needs to be a very personal conversation and it saves the clinicians time from doing the data search to be able to have that conversation about, ‘Here’s some options for you, here’s the pros, here’s the cons,’ really wanting to find the best two or three targeted options that are specific to that patient’s genomic sequencing and leveraging the body of data. Not just the data that that person or that organization might have access to, but really leveraging the body of information that’s out there in the world on this particular cancer type and surface relevant options for that particular patient.

It’s a continually evolving clinical field, like most of them are, but these newer, cutting edge ones are evolving more quickly. So, it’s all the more important to have technology that helps support those decision makers and augment their intelligence with things that may have just been published or just been discovered. Things that maybe are uncommon or rare often happen in cancer patients as well, where they’ve got a particularly rare marker and so it’s not that the doctor or the decision maker isn’t well versed on everything, but this thing is so rare that perhaps they haven’t seen it in their practice. All the more reason to have technology be able to connect those dots and find those patterns from the specific sequence of that patient and then across the data sets available in the body of literature.

VN: Where do you see AI in healthcare going next? What technologies are on the horizon that will get it there?

TM: Great question, and if I had a crystal ball or a Magic 8 Ball to answer this question it would be fantastic.

I think it will be a fascinating process and evolution to watch. Certainly, the growth of data is not going anywhere in our world in general, and it's certainly not going anywhere in healthcare. While that’s overwhelming, it’s ultimately a good thing for the industry of healthcare. The more you know, right?

The opportunity, then, is to continue to evolve the technology to be able to help regular old human beings be able to make sense of that vast amount of data. That dynamic is what will continue to drive and push the technological advances in healthcare through both analytics and the advanced analytics capabilities of AI. Sometimes you need one and sometimes you need the other, and not everything needs to be AI. That means finding that balance of where good old fashioned advanced analytics, which are still very powerful tools, and we have many of those solutions as well, finding that balance of how draw out of data the information that a decision maker needs, wherever that data might be coming from, whatever it might look like. How do we support a human being’s thinking by pulling those things together in a way that we do as human beings, when we look at a variety of data and we make a decision? When you’re driving, you’re taking in a variety of data points as you’re driving and making decisions so that you get to when you’re going safely and as quickly as possible. How do we build technology that has those same connection points and identifies those same things that a human would do if they had the ability to read at the scale and the speed that AI does and surface that up to them in the most effective way? That’s whats driving the technology, that’s where we will continue to focus our efforts to really make a difference in how healthcare is practiced, but also how the health of individuals is continued to be improved. So, things like social determinants, where we live, the kinds of foods that we eat, all of the things that go into a person’s overall health, which, again, is largely not just the healthcare component. That’s important, for sure, but 70 percent of it is stuff in addition to that. How can we apply AI to make sense of all of that and help each patient find their best path?

VN: Is there anything else I should know?

TM: I think the thing I would leave you with is this idea of our vision being how do we use technology for the good of mankind? How do we have technology supporting human decision makers whose job and calling is to help take care of patients and help patients take care of themselves, as opposed the system making the decisions for someone, or replace roles, or anything like that? It’s really about how do we help human beings address the challenges that the industry us trying to address: reducing variations in care and rooting out waste and improving patient and improving outcomes. We have the challenges before us that that humans have before them and our just is to supply technology to help them effectively impact those challenges. It’s a pretty cool industry, for sure.

(The Meet the Corporate Innovator series is brought to you by Advsr, a startup advisory firm in the business of starting conversations and sparking big ideas.) 

Support VatorNews by Donating

Read more from our "Meet the Corporate Innovators" series

More episodes