English articles
Are data taking over and do humans take benefits?
Panel discussion at Data Natives conference 2019
At the last Data Natives conference, antwerpes CEO Thilo Kölzer moderated a healthcare-related panel discussion on Big Data, AI, Deep Medicine and DTx. Nicole Büttner, Dr. Carina Walter and Dr. Roman Rittweger were the panellists.
Thilo Kölzer (TK): Welcome to our panel with the title „AI, Deep Medicine, DTx – Are data taking over and do humans take benefits?” This is the topic I would like to discus with my panellists today. First of all, I would like to ask you to introduce yourself and tell us what you’re doing in your daily lives and which are your touch points regarding data and healthcare. Nicole, please start!
Nicole Büttner (NB): Thank you very much and thanks to all of you for joining us. My name is Nicole and I’m the CEO of Merantix Labs. I spend my day creating machine learning solutions for clients in a variety of sectors, including the healthcare sector. I’m part of a machine learning and AI-based company builder here in Berlin that’s called Merantix. Within the umbrella of our venture studio, we also have a healthcare dedicated venture which is Merantix Healthcare and which has a certified medical product in the market in the area of breast cancer screening. The product is called Vara and is a workflow automation tool for radiologists to interpret mammograms. This is one big touchpoint for Merantix and specifically at Merantix Labs we work with insurance companies, hospitals, pharmaceutical companies and laboratories – so quiet varied interactions.
TK: Thanks! Roman, do you want to continue?
Roman Rittweger (RR): I’m Roman and I’m a physician by training. I worked at a consultancy and at a healthcare start-up service provider for health insurances. And four years ago, together with two co-founders, I started ottonova, the first digital health insurance in Germany. Data is very important to us and obviously, we use data everywhere. We started now in the digital area so we’re using data when we try to acquire new customers. We’re using data to keep our customers happy and we’re using data to keep claims costs low. And to keep our customers health insurance costs low over time. So it’s a very important issue for us.
TK: Thank you Roman. And now, Carina.
Carina Walter (CW): My name is Carina, I’m a computer scientist and I started at Boehringer Ingelheim two years ago as a Data scientist. In our team, we develop data strategies, to enrich data and ensure data validity, actuality, as well as quality. At BI, many diverse data sources exist and we do not want to store them in silos. Further, we calculate prediction models to identify customer needs, e.g. to support HCPs.
TK: OK, thank you all! So up on stage, we have a start-up company in healthcare, an insurance company called ottonova and a doctor. So you’re the only real doctor here on stage, Roman. And with Boehringer Ingelheim, we have big pharma as a guest. So let’s start, dear panellists. I recently read a book called “Deep medicine”, it’s written by a physician called Eric Topol. And I’ve read that the annual amount of data produced per individual is about 1 Terabyte. That’s one Trillion Bytes. And I think we all know there are more data floods to come, especially regarding health data. Compared to the last decades, the data sets nowadays are more digital for sure, exponentially bigger and richer and matched by remarkably computing power and algorithmic processing.
Compared to the last decades, the data sets nowadays are more digital for sure, exponentially bigger and richer and matched by remarkably computing power and algorithmic processing.Thilo Kölzer, CEO antwerpes ag
So Roman, the first question to you. And I will start with the following quote: Five years ago, Tim Harford from the Financial Times wrote in an article that said “Big Data has arrived, but big insights have not”. Would you still agree on this statement or what do you think what changed in the last five years?
RR: I would like to give you two answers: First, the physician answer and then the ottonova health insurance answer. The physician answer is: I would say yes, we see some first results of big data. But we’re not quite good at big data to be honest. To use big data in a good way, we first have to use the structured data in the right way. We must take this into account. I think we have lots of rule-based data management ideas. Even behind the app Ada here in Berlin there are many rules. Nevertheless, I think that’s the beginning for using big data. And in health insurance, we’re using data when we’re looking at claims and try to see whether the claims are correct or whether somebody is trying to defraud the health insurance. And that’s really good. But even there, the current systems are actually rule-based and not using artificial intelligence, the way that we all think that we’re using it.
TK: Ok. So, regarding your physicians’ point of view, you don’t see that we use all this data adequately so far. But regarding ottonova or insurance business in general, you use data to manage your members for example.
RR: I guess the area where we use the most AI and big data is when we buy AdWords at Google. Then Google is using it for us.
TK: All right, so I can understand that a doctor feels like to be threatened by technological progress – but I also see the opportunity to turn the tables. All these apps, which are currently published, like Ada for example, are not reserved only for patients – doctors can use them to improve their work and to improve their output. They have lots of possibilities to extend their capabilities. Nicole, would you agree to the statement that if you want to satisfy curiosity, you use an app, and if you want to get answers, you consult a doctor?
NB: Wow, that’s almost a philosophical question. I think there is actually a gap between those two. So for example I think to get really powerful insights, it would be great to link all the apps I’m using to the doctor I’m consulting. Because I’m using all these great apps about how much I sleep, what I eat, where I travel, how much sports I do. And basically my device almost knows more about my daily routines, my resting heart rate etc. than my physician. To me, it would be really interesting to provide this data to my medical doctor in order to give her a more complete picture of my actual health state and may inform a more comprehensive treatment path. And I think that’s what’s exciting – that we’re moving to an area where I can get lots of transparency and information about my own health state and it’s becoming more comprehensive. And I think we need to make that link to really create these powerful insights, because at the moment it’s just a part of the big picture when I physically go to my physician.
I think that’s what’s exciting – that we’re moving to an area where I can get lots of transparency and information about my own health state and it’s becoming more comprehensive.Nicole Büttner, CEO Merantix Labs
TK: You say that you have wearables that track your health data. All these data is stored in different places, and different apps and different databases – Do you know a doctor who could use your data?
NB: I think honestly the most digital interaction with my doctor using sort of “digital data” is when I took a picture of my suture after an operation and he said, “Wait another two days to change the suture”. But we all know – I mean we’re sitting here at Data Natives Conference – technically it would be possible. I think our collective ambition should be to move to a model where I still have control over my data, but where I can easily make it available to my physician for consultation. If there’s the possibility, I think let’s do it.
CW: A comment on that: I think the biggest disadvantage is that data is stored in different places. An integrated data solution would be desirable. Of course, this procedure has to be GDPR compliant. Everyone still has to have the rights on their own data, and has to be willing to share it. Based on such an infrastructure HCPs would probably use such apps. However, a HCP has no time during the daily business to check diverse data sources to get a completely big picture for one patient. That is why I guess building up such a GDPR compliant infrastructure has to be the first step here.
I think the biggest disadvantage is that data is stored in different places. An integrated data solution would be desirable.Dr. Carina Walter, Senior Data Scientist Boehringer Ingelheim GmbH
TK: Let’s imagine you have twenty patients sitting in a medical practice and twenty patients would use twenty different apps. The doctor has no chance to look at all these data collectors.
RR: Even if they would use all the same apps, it is still a physician’s nightmare. Because he would spend more time than he did before. And he wouldn’t get any more money. So you have to get into the insights of a physician’s head. Currently you come into his office and you listen to him for one minute and thirty seconds and then he gives you a prescription and you’re out again. Imagine you bother him with all the results of all your different apps. It’s a physician’s nightmare.
TK: That’s a good point. But lets say data and digitization promise to automate diagnosis, and make it easier and automate care. And at first glance it seems to be obvious to replace humanity by technology. But what do you think when it comes to concrete patient care and the last mile, meaning it the talk between physician and patient, meaning the human interaction in the end. Should that remain analogue or do you see a chance to digitize it? How important is the human touch in the end?
CW: I think the personal interaction in patient care is necessary and even in times of digitization; it always has to be there. However, you can use all the data and the digital transformation in a first step, to be able to generate a more precise and holistic picture of patients. Based on the complete information, the physician is able to support and accompany patients in a more targeted way, considering all patient needs. Because of the entire digitalized world and all the data which is available you can get an insightful picture of disease patterns and a more focussed view on what patients need. In my opinion, the physical interaction has to be there, but using all the data and analysis, we are able to support HCPs to treat patients more precisely and on an individual basis.
TK: What does a pharmaceutical company like Boheringer do to empower the physician to get these concrete data and to get the big picture to handle a patient?
CW: We at BI try to support HCPs in getting precise and accurate information about disease patterns, he or she actually is interested in, without losing valuable time by searching through lots of websites. The benefit is: HCPs have more time for treating patients and they get more relevant information faster.
TK: So your aim is to empower HCPs to work with you therapies.
CW: Yes, that’s right.
TK: Nicole, at Merantix you’ve developed an image recognition software called Vara which helps radiologists to diagnose breast cancer at a very early stage. What’s your promise regarding Vara?
NB: One of the challenges is (luckily!) low incidence. Just to give you a few numbers on breast cancer: From one thousand women who go into breast cancer screening, under five per cent have to go to a second screening. And about six to seven are diagnosed with actual breast cancer. 20-30% of breast cancers are overlooked, so not diagnosed in either of those two screenings. That’s why the goal and the mission are to empower radiologists to help addressing this problem. And we think this is due to different facts. At times, physicians are under time pressure, given on average about 40-60 seconds per mammography case. Additionally, we know that some cancers are ‘occult’ and not even visible to the human eye at the time of screening. We believe the major challenge is the low incident setting though. The software product Vara that Merantix Healthcare has developed is basically a workflow automation tool for radiologists. It is a smart classification system that looks at mammograms and automatically classifies all the ones that are cancer free. The rationale is to really empower radiologists and the issue with low incidence of breast cancer is: When you don’t see things often, you often don’t see things, right? So Vara shows the radiologists substantially less cancer free mammograms, which increases relative incidence in the remaining mammograms. Currently, 40% of cases can be automatically classified with Vara in this way. Vara pre-fills the medical reports for those, which the radiologist only has to proofread and send off. That’s a huge time saving for the radiologist and he or she can focus on those cases where it might be more ambiguous. So that’s the promise. And we obviously hope that we will be able to increase this number successively over the next years.
The issue with low incidence of breast cancer is: When you don’t see things often, you often don’t see things, right?Nicole Büttner, CEO Merantix Labs
TK: So for the moment, Vara focuses on breast cancer screenings. Do we have other indications in that area that could be interesting for Vara in the future?
NB: My understanding is that my colleagues at Merantix Healthcare want to master this task very well. From 40%, which already adds a lot of value, there is still room to improve. I think their first priority is to really deepen the performance on breast cancer cases. So, fully understanding breast cancer with different diagnostic tools, before moving to other medical imaging cases like chest- or head-CT-scans. But what the exact roadmap looks like is a question for Jonas Muff and his team. And remember, it’s only just been CE-certified which means it can only now commercialise this product and roll it out clinically.
TK: All right, so you’re certified and you’re already on the market with Vara?
NB: Now Vara can be commercialized, yes. The process of certification took one and a half years first for Merantix Healthcare to become a certified medical device manufacturer and then to certify the medical product.
TK: So Vara is a medical device. Roman, as you’re the CEO of ottonova, what’s your opinion about these digital medical devices, which have come up? It’s certified, so your members can use it.
RR: I think we’re drilling the tunnel from two sides of the mountain. We’re coming from the payer side and we’re starting in the first step by giving our members digital access to specific knowledge to find out which kind of preventative care is the right thing for me right now. And when they go and have this preventative care we give them a checklist of things a physician should do. So basically we try to make it easier for the physician, so no extra work, but also make sure he does all the right things. And we also have a digital physician visit, so you can just video call from your phone and get a sick note. For instance, my son the other week got one so we didn’t have to go to the physician and sit in the waiting room. You know, all of our insured lives send their bills to us via their smartphone and we have the data right away. So in the long run, we’ll be able to see from the data that would be right for whom and what kind of measure. We could invite them to take digital visit of some kind and we could also pay for that. And we could have an additional payment if it works well and we see that the insured person is staying well afterwards. So we have the infrastructure and place for this but we’re just starting with baby steps on our side.
Usually that’s the problem that a normal health insurance has. They don’t know what the patient needs at what time. They just wait for the bills.Dr. Roman Rittweger, Founder & CEO ottonova Krankenversicherung
TK: So you’re ready to support doctors that use software tools like Vara or other apps for example, right?
RR: Yes. We could even funnel patients to them. Usually that’s the problem that a normal health insurance has. They don’t know what the patient needs at what time. They just wait for the bills. But we have the data and we could actually send the messages and say, “You should be in line for this preventative treatment now and that’s how you can get it”.
NB: And I think, if I may just jump in here, this is fairly critical when we look at deploying these technologies in the healthcare sector. Because there will be some sort of tech-dividend or digital-dividend and sense of a benefit that’s created, right? I think we’re all working hard for this to be to the benefit of the physician who’s treating and the patient to have a better, more meaningful interaction and better quality of treatment. And of course, at the same time, we’re all looking at how high our healthcare expenses as economies are and can we help reduce them while improving quality of care. In my personal opinion those are two factors to weigh and I think for us, to be successful, and maybe potentially for all of you, we have to keep this balance in a good place so that the patients feel that this does not only cut costs for some hospital or insurance company, but also improves their health outcome and their healthcare.
RR: We always need a win-win-win. The patient needs to win, the physician needs to win and the health insurance needs to win because in the end, that’s what the patients will have to pay for in the future.
TK: So coming back to the data aggregation-part from the beginning. I think you as an insurance company are in a good position to be a dig data aggregator and to use this data for the members.
RR: We think that we’re the paying layer of the healthcare system. We pay for the party. We’re also getting most of the data, especially as a private health insurance because we get all the bills. It’s more difficult with GKV in Germany (statutory health insurance), so we are in a great position to do this. And we’re also acting in the interest of the patients.
TK: Will you do this or are you already doing this?
RR: We’re setting the basis for this and we’re even selling part of our software to other health insurances so that will also be able to do that. We’re growing so fast that we can cover all of Germany in ten years.
TK: All right, good! And Carina, I think there’s this fourth win, that’s the pharmaceutical industry. So what does Boehringer Ingelheim, one of the world’s largest family-owned pharma companies, do, to combine data, big data science and drug development?
CW: Boehringer Ingelheim is investing a lot in building up internal capacities. We have a data science team in our global department and we have a digital lab, BI X. At BI X, 50 people are working on diverse projects using digital innovation to develop digital healthcare solutions. The digital lab works closely with the business partners at BI to do very innovative cutting edge projects. I think this mind-set-change is the first step you have to take as such a big company. Further, while working with data, it is important to ensure data quality and to have a GDPR compliant infrastructure to store data. In order to do so, we are using a global BI data lake.
TK: You mentioned that Boehringer empowers doctors. Do you also want to improve the patients’ quality of life with the help of data or apps or digital treatments?
CW: Yes, there are several projects at Boehringer Ingelheim to improve patients’ quality of life. A group at BI X is doing research in speech recognition, to be able to diagnose Alzheimer patients in an early stage, because it is known that language changes when having this disease. Further, Boehringer Ingelheim has the Angels Initiative. This is a unique healthcare initiative to improve acute stroke care. They help hospitals to become stroke-ready, to be able to treat stroke patients as quickly and effectively as possible. An optimal acute stroke care can save lives and disabilities can be prevented. Thus, the initiative build acute stroke networks to optimise treatment and diagnosis as well as implement best practices.
TK: So there are many different initiatives. Roman, do you appreciate such kind of initiatives from the pharmaceutical industry from an insurance company’s point of view?
RR: Yes, I think it’s great. Quite frankly for our members it’s not a big issue yet as we’re insuring at this moment young healthy lives. So our main thing that we’re trying right now is the “eat, move, mind”: How to eat well, how to move more and how to relax. So we’re supporting our young healthy members with that. But as our population will age over time, we’ll certainly move into these areas and we lay the foundations for that.
TK: So you will probably find new members at Merantix Labs for ottonova (laughs).
NB: Who knows? (laughs)
TK: Carina, coming back to the new competitive situation for pharmaceutical companies. Let’s talk about DTx, the digital therapeutics area. New digital treatment apps and digital initiatives come up. There’s a company called Voluntis for example that is active in diabetes and oncology. And I recently read that Google is a pharmaceutical company now because of a new pharmaceutical division from Google: Verily Life Sciences. How do you look at these new initiatives and these big players like Google in your core-market?
CW: I think, what Google and other big tech companies are pretty good at is to handle big data well and get useful insights out of it. I would not say that we have to be afraid of Google but I think we have to have them in mind and to learn from them, for example, how they integrate all the insights and how they build up the big picture.
I would not say that we have to be afraid of Google but I think we have to have them in mind and to learn from them, for example, how they integrate all the insights and how they build up the big picture.Dr. Carina Walter, Senior Data Scientist Boehringer Ingelheim GmbH
TK: Roman, can you imagine that so far non-pharmaceutical companies outperform the established industry players?
RR: In the long run it could very well happen. Just look in the past: The best business model was the pharma business. Why? There’s always a therapy, you can always pop up a pill and you can always change your behaviour. Changing your behaviour is very important, but it’s very hard to prescribe and to do to be successful. Popping a pill is actually very easy to prescribe and very easy to follow through. Even there are some issues with compliance and adherence, but we’re getting better in changing the behaviour with telling me how many steps I did through the day and sending me reminders for instance. That is going to get much more important.
And the question is: Which one of these things are the ones that are keeping me at a better health? And that’s what we’re doing at ottonova right now: We’re giving money to all of our customers if they want to buy the Apple Watch or want to do something in that respect because we know that gets them healthier. The thing is: If I have a specific disease, how do I put these algorithms to work on that specific disease? That’s something where you need knowledge and knowhow what the tech-companies don’t have yet and the pharma industry would have, so the question is: Who is going to grab the territory?
TK: From a doctor’s point of view: How sceptical are you regarding the prescription of such kind of digital treatments? There’s a new law in Germany: From next year on, you can describe digital treatments or apps. But I think there’s much scepticism, right?
RR: I think in general it’s a great idea because it’s actually very cost-effective and can be easily administrated and with our platform I think were in a great position to do that. What makes it a bit more complicated obviously is, that we don’t have any studies proving that an app is beneficial. Because you know when you program the app and when you’re living in the app-world-tech-industry you know you change it every three months. But to get the studies in place to see whether it’s working, it would take three years. The main issue is here: How do you know whether the app is working or not. If its cost-effective or cheap enough, you can basically just let it run three years and find out whether it’s working or not. That’s the challenge really.
TK: I think we’re running out of time so we have to come to our final round and I ask all of you for a short answer for my last question: From your point of view, what kind of dividend can be received by investing in data science and medicine?
NB: Well, I think we can improve in medicine, right? And we can basically scale good solutions more efficiently to broad populations. In the long run, it will therefore be a driver for inclusion and availability of the best care standard to patients. And I am convinced this can lead to more personalized and human interactions with patients and will support professionals who are working in the sector to lift up the important work they are doing.
TK: So to win more time for the human interaction in the end.
NB: Yes, for the benefit of a better patient care.
RR: I think we can get the knowledge personalized to the end user. That’s the biggest digital debit that I have in my mind.
CW: I agree. I think this is one major benefit, as well as a faster development. Using integrated data and generating insights out of it can lead to a more efficient and goal oriented research.
TK: Okay. All right! Thanks a lot for sharing your thoughts with us. And thanks to you, the audience, for stopping by.
Author
