The New Era of AI-Supported Health Care
[AUDIO LOGO] JENS DOMMEL: What we are seeing is more and believing more that AI becomes a new member of a health team in a hospital, for instance, but also an ally for the person at home.
ALEX MAIERSPERGER: Today, we welcome Jens Dommel, head of health for EMEA at AWS. He's going to tell you exactly why you need AI on your healthcare team, and he'll share his favorite stories of innovation from all around the globe.
[MUSIC PLAYING]
Jens, welcome to The Health Pulse podcast. Thank you so much for being here. We're excited for this conversation.
JENS DOMMEL: Pleasure for me as well. Thank you for inviting me.
ALEX MAIERSPERGER: Jens, when we think about Amazon in the US, we think about the packages that get delivered to our door, and so supply chain and logistics. When we think about Amazon in healthcare, we think about the acquisitions and integrations of One Medical for primary care, PillPack on the pharmacy side. And then when we think about cloud computing, we think about the AWS side. How should we think about AWS in EMEA in healthcare?
JENS DOMMEL: Yeah, sure. And you named a couple of services, and it shows that we, as Amazon, we're focusing on the problem-solving. Wherever we identify the problem is saying, let's think about how we can solve it and help it.
And you named a couple of services from Amazon, like One Medical. It's a primary care service to help-- to do-- drive and deliver virtual care anywhere. As an example, or Amazon Pharmacy, it's an American service for pharma delivery at home, delivering prescription at home.
And when we are talking about AWS, we are the IT arm of Amazon. That means we are focusing to delivering a service and infrastructure worldwide-- high-scalable, secure, affordable, and reliable, though because this is what's needed when we talk about digitalization transformation, that you have a secure environment to run your application, your services where needed.
And we are talking-- and I'm responsible for the healthcare division in Europe, Middle East, and Africa. And what we are seeing that this kind of digitalization and transforming the health system is key, especially when it comes to the challenges we see and we are facing in the healthcare systems.
ALEX MAIERSPERGER: Jens, you mentioned all of the countries that you serve and on multiple continents and places and all of your travels. How are the-- are there similarities in the challenges that you're seeing within those countries, or is it very widespread? You obviously have to point this technology towards ministries of health, towards individual patients, physicians. What are the challenges that you're seeing that may be similar across these geographies?
JENS DOMMEL: Yeah, Alex. This is a good question. There are similarities, but we are seeing also differences where we have mature health systems, as well as like in Africa, we are talking about access to health. The health equity is a big topic there.
But overall, we are seeing-- we are in an aging population in 2020 with the Boomer, and we have above 1 billion people older than 60 years old. And the projection is till 2050, that we are talking about 2.1 billion. And the average life expectation is 79 in US and 81 in Europe, for instance, and it's growing as well.
So if we are seeing at the other hand, the challenges in the health systems, there are a lot of those. Like nearly 1 in 4 admitted patients in a hospital suffering adverse event. And the primary care is 4 in 10 cases, as an example. And 50% of those are preventable.
There are 10 million challenges in the diagnose errors. Imagine we can manage them better, as well as the cost is a bigger challenge because this is limited, especially in the public sector.
Overall, we are seeing a gap in the clinicians. So WHO project 10,000,000 in 2030. And besides this gap and the shortage, there is also a challenge with the burnout in this area. It's growing very, very dramatically.
So this is a challenge. If you bring all together, it's a question of how you can solve it. And we see here that AI is one of those pillars and technologies where there's a big potential in it.
And McKinsey, for instance, has shared a study showing that there are potential of 30 to 40 better patient outcome and better processes and reduction of the patient costs by 50%. So this is where we see also that technology can help overcome those challenges.
ALEX MAIERSPERGER: You mentioned all the heavy hitters on the challenge side of aging populations. And I think we're seeing that pretty much all over the world. Populations are getting older. They're also staying alive longer while sicker. And so we're getting older, sicker. You talked about physician burnout and clinician burnout. And so there's workforce challenges.
And then you mentioned AI. And so there's some that feel like-- I think we see some of those headlines in the news that AI is going to replace physicians, or AI is going to deliver care. There's very sort of audacious statements around AI. And then there's more tempered thought process of AI is not going to be the thing that fixes it.
How do you-- you mentioned AI within those challenges. How do you see specifically AI being able to deliver health care access and affordability and opportunity?
JENS DOMMEL: Yeah, yeah. So the AI-- and you mentioned a couple of these fears around AI, substitution everything, and being self-made like this. What we are seeing is more and believing more that AI becomes a new member of a healthcare team in a hospital, for instance, but also an ally for the person at home.
So whenever there is a challenge that you have a source which can help you with some guidance, which you can ask and get an answer. But at the end, the doctors is doing the decision. Or me, as a person, I'm still doing the decision.
What's needed, and we call it democratization of AI, is an access to those technologies. And we, as AWS, our target is that every company, every institution become an AI organization. And if you think about what needs to be delivered, it needs-- one, you need the easiness of use of AI services, or that you are able-- that everyone is able to build an AI application at once.
At the other side, AI is compute hungry, and you need the scalable infrastructure, global available. This is what we are delivering, for instance, and what we are investing heavily to make it much more available across the world.
And that this is another thing. The other is in the health care environment, you need to train your models, your AI models. We are seeing a lot of investment in it, and also the trust in the data and the reduction of the hallucination.
Because if you ask an AI system today or, for instance, a GenAI, maybe you've seen it using GPT, ChatGPT, or other-- Qure or others-- you always get an answer. But in the healthcare space, you need to trust the answer. They need to be predictable.
And this is where we need special services, trained services, maybe with your own data. And doing so, you'd need a secure infrastructure. That means when you put data in yourself, maybe patient data, you need to be secure that those datas are secured and not going outside of your control. And this is where we are investing in, and with our infrastructure, with Bedrock, SageMaker to provide those services to support this.
And at the end it is about cost. If you think about AI at scale, and you think about what is the cost of a prompt, and we are seeing there is a big demand to reducing it, those costs of prompts. That's why we're investing not just in services to build and use application, but also in infrastructure services, like chips, like the Instance, like Inferentia, for instance. This is where we believe that all impacted facilities needs to be considered to deliver those to get the best out for the healthcare system.
ALEX MAIERSPERGER: You talked about trust in healthcare, how it's not just about asking a question and getting an answer, but it's about getting an answer you can trust. And then you said, it starts with the data. Can you tell me more about that?
JENS DOMMEL: Alex, It's a fundament because the output-- the quality of the output belongs to the quality of the input, and the input is data. And what we are seeing today is that we're coming from an application-centered infrastructure where we have the hospital information system, the PACS system, the labor information system, the ERP system, and so on and so on.
And this disadvantage here is that the data are siloed very often and not good accessible. And we are seeing now the trend in building those health lakes, multi-modal health lakes or data lakes, where I see especially our cooperation with SAS and AWS is really relevant because we are unlocking, we are unsiloing those data, make them accessible, building a governance model on top of it, but also having, with the scalability of the technology and the compute power, the capability to do a good research analytics. And this globally. And this is where we need this kind of technology to provide the foundation building the right AI model on top of it, but also turning data into real value.
ALEX MAIERSPERGER: You talked about AI teammate and AI organization-- really exciting vision here. And then you mentioned both cost-- and so figuring out the cost per search and query-- as well as the ability to pull the information when you need it and where you need it across various geographies and things-- so data sharing and privacy.
And love how you talked about trust. Healthcare is probably one of the highest trust industries. And so being able to say, this is information that I need that I can trust, that comes from the data that we understand where it's coming from and how it's coming from-- really exciting vision that you talked about there.
You also talked about-- right before that, you had talked about the maturity of health care delivery organizations or financing mechanisms across your leadership. Can you tell me about the transformation process?
And so obviously, there's organizations, health care organizations that are more ready to have an AI teammate versus be a full AI-enabled organization. How do you go about getting someone maybe in the middle of that curve all the way to the head of that curve?
JENS DOMMEL: [LAUGHS] Yeah, it's a big topic. And it's not just AI, and it's not just technology. It's about the combination between organization processes and technology-- how you bring those together, in a way. And most important here are the workforce-- the skills of the workforce and the motivation and the culture in the organization.
What we are seeing were companies, organizations delivering really well is whenever they have a strong leadership in setting really stretch goals at once. But at the other hand, also not boil the ocean. Start with a small projects, learning how it works and then scaling.
At the other hand, if we are seeing that on a national level, the ones who are really making good progress, they have-- most of them have a cloud-first policy. So they are challenging themselves to say-- not that it means that everything needs in the cloud. They are saying, OK, if we have a new project, you argue why we don't do it in a cloud, instead of the opposite is one second.
They are building special frameworks. Because important is also to simplify the process of the transformation is to define frameworks for architecture, for security, for procurement. And architecture, for instance, there is a difference if you're talking about data management, architect and security about data, they are different clusters. Not every data set is a patient data and needs to be the high level of security and compliance, where, at the other hand, there are administrative data.
So that means this is something what you can balance well and where you can need different services and solution and infrastructure. Not to forget the procurement because cloud is delivering a new capability to deliver services.
In the past, we had tenders. And tomorrow we have tenders. But the way of setting up a tender which is cloud-friendly looks different than the ones on-prem. In the past, they defined every single function and quality criteria they want.
In the future, and with cloud-friendly, what their customers and partners are doing is more defining what they want, the Service Level Agreements or the SLAs, and let the technology deliver the best solution. Because the innovation is so fast, that it's so challenging to define especially which function from which cloud service you need. Let the service teams doing it.
By the way, there is a good source called CISPA. It's an independent org, which giving good guidance for government, especially how to build such a procurement framework. And what we shouldn't forget also is there are special marketplaces now where we have a broad set of partner solutions in a marketplace.
So that means if you really want to use it and easy to deploy it, you can do it by click. And in a couple of minutes, you have a running system in your environment. But therefore, especially when it comes to public sector environments, you need the procurement frameworks which allowed this. And this is something what is good described also in the CISPA how to do it and how to enable it so that the customers can use and get the best out of the technology innovation.
ALEX MAIERSPERGER: I really like how you laid out that maturity curve. And so on the lowest end of it, or if you haven't started, I love how you said just start. Pick a project and start to dabble with it, play with AI, be able to really see its capabilities.
And then on that other end, if you're a mature organization, starting to understand the governance of where to use and how to use patient data verse administrative data and the capabilities that exist, that you really laid out a great maturity curve of-- and I think people listening, if you feel like you're somewhere in that low end of the AI adoption curve, you can say, OK, we need to pick a project, and where are these things? So really, really great answer.
All right. You've made it to our speed round. And so we're going to come fast and furious with some questions. And we want of yes, no-- quick answers. What is the most used app on your phone?
JENS DOMMEL: Let me think. For sure, my email, WhatsApp, and LinkedIn.
ALEX MAIERSPERGER: All right. That might be too much work. Let's see. Let's say what's the most used non-work app on your phone?
JENS DOMMEL: Non-work app? Let me--
ALEX MAIERSPERGER: Leisure app.
JENS DOMMEL: I think it's a weather app.
ALEX MAIERSPERGER: Weather? OK. That's important for all your travels. What is your--
JENS DOMMEL: Always.
ALEX MAIERSPERGER: What's your favorite type of exercise?
JENS DOMMEL: Oh, I wish I could do more, but in my free time I really like the ball sports and like to play table tennis or volleyball.
ALEX MAIERSPERGER: All right. Table tennis. That's an-- OK. That's an exciting one, one we haven't heard as much.
You cover a lot of countries in your day job. What's one you haven't visited yet but want to?
JENS DOMMEL: Oh, I guess there are more that I haven't visited yet. I haven't been in the Asia area. So like Japan, for instance. I think this would be a nice experience, as well as Australia or New Zealand is also a great place to be. I hear a lot of the innovation they are driving in healthcare, by the way.
ALEX MAIERSPERGER: Exciting. Your favorite dessert. So we're going far away from the type of exercise to dessert.
JENS DOMMEL: Oh, ice cream. Vanilla ice cream. It's my favorite. [LAUGHS]
ALEX MAIERSPERGER: Vanilla ice cream. All right. We're kindred spirits. I'm an ice cream person at heart.
Do you prefer mountain or beach?
JENS DOMMEL: Beach. So that you see, I'm the-- I prefer more the warmer area instead of the cold winter time area.
ALEX MAIERSPERGER: Yeah, this is-- ice cream at the beach. I think you've got this mapped out.
JENS DOMMEL: Imagine what could be better, yeah?
ALEX MAIERSPERGER: How about, do you prefer morning or night?
JENS DOMMEL: I'm a morning person. I like to get up and see the awakening of the day with the birds chirping. And best would be sunrise. So I like that really much. Yeah.
ALEX MAIERSPERGER: All right. Well, appreciate you going through the speed round. Exciting to get a little bit of glimpse into your personal life and how you operate.
One question on the regulatory side. There's a statistic that almost half of the world's population experienced elections last year in over 60 countries. And so obviously, at the forefront of what the potential for AI has is how the regulatory bodies are reacting to it and either allowing it, helping it, propelling it, or maybe limiting it.
The European Health Data Space obviously has the EU AI Act. What changes have you seen in the market because of that. Are countries feeling excited by this? Are they limited by this? Are they finding ways to play within these boundaries? Is this what they want?
JENS DOMMEL: Yeah. It is always a question how much regulations you need, that it's not limit the innovation. At the other hand, that you can trust an environment. And this is what we are saying AI for good, that it's used for good and not for the bad. So this is important that there are some regulatory borders like the AI act in Europe.
And the European Health Data Space, there are special made to unlock the potential of data and to improve the cooperation between the countries, but also within the countries. So a common standard is something what is helping.
Because I've been now since 20-plus years in the health space, and I've started-- I've heard interoperability challenges 20 years ago, and we hear the same today, so that there are datas but they are not-- can be processed because they have different formats and so on. So that's why those kind of standardization brings value accelerate processes. And European Health Data Space is one of those initiatives.
And we are committed, by the way, to the EHDS. We have implemented a special initiative with the single threaded owner. We call it AWS for EHDS. And we are providing reference architecture showing how you faster can build solution.
For instance, for secondary use, we have a reference architecture in our portal. Or also, we are combining our ecosystem. AWS is a partner-first organization. That means we have partners building their products and application for the healthcare industry, as well as in others, other industries as well for sure.
But nevertheless, coming back to EHDS, this is where we are seeing that this partner ecosystem, providing different services in building those core infrastructures or data spaces, as well as in serving-- putting additional services on top of it. For instance, for medical research, for radiology, for instance.
Or if it comes in a OMOP, in a clinical trial environment with special OMOP service, to name here some of those standards where we are working together with InterSystems, for instance, or with Xavier or other comparable partners.
So this is where we also see that EHDS, building the foundation, and it make us easier to bring the partners together, building a comprehensive portfolio to drive the strategy to turn data into real value and to unlock the potential of data and to avoid the silos, what we are seeing today.
ALEX MAIERSPERGER: I've long had a saying that the future of health is together. And it sounds like this partnership, strategy, and opportunity is very much in line with that, of it's going to take all of us to get to a healthier future. So really exciting to hear.
JENS DOMMEL: Indeed, it's a team sport. Yeah. No one alone can solve this. This is something what we need to do together with all the different advantages, skills, and capabilities.
ALEX MAIERSPERGER: You've seen and pinpointed some of the real challenges of access in rural areas, and maybe access in other areas as well-- the aging workforce, workforce challenges, burnout-- all of the things that we talked about from a challenge perspective, and you've seen them very acutely. What makes you optimistic that we're going to get to a healthier future?
JENS DOMMEL: I think the one-- I've seen a lot of passion person in the healthcare space. And the beauty and the challenge on this industry is we're talking about here about human lives. And this is, by the way, also my personal passion and motivation to make the systems better globally. And I'm seeing a lot of my friends, colleagues, and partners, they are-- this is a common motivation. And this gives me one of the trust that we are able to do it.
The second is that with the technology now, we can overcome problems faster. Imagine today, 97% of the data are unstructured, and data is a foundation for AI. And AI can help to turn the unstructured data into structured so that become better data sets foundation to make-- to support the diagnosis, to support the research, to support the doctors and nurses.
This is something where I believe also that the technology and the capability of the technology plays a big role. And that's why I also believe that we're going to see progress in the better outcome, better quality in patient safety.
ALEX MAIERSPERGER: Yeah, I love how you immediately brought it to the people, of-- even as the technology of the people inspire you, and the technology helping us get there faster is so exciting.
Jens, thank you so much for joining. I feel like I need an honorary-- or I feel like I get an honorary Global MBA from this conversation of hearing how the world is using data and AI to drive this healthier future. So exciting. Really appreciate you taking the time today.
JENS DOMMEL: Thank you, Alex. Great question. Looking forward to talking to you soon. Bye-bye.
[MUSIC PLAYING]
ALEX MAIERSPERGER: Thanks for listening. If you have questions, or if you want to join as a guest, please email us-- thehealthpulsepodcast@sas.com. See you next time.
[MUSIC PLAYING]
Jens Dommel, head of healthcare for EMEA at AWS. He-- [FLUBS LINE] [BLEEP] That was some enthusiasm. Is that better? Yeah, I think that was-- I think-- no, I think that was pretty good. Got him.
Creators and Guests
