Implementation & Internal Innovation: How the Largest Challenges in Hospitals Are Being Addressed by Emerging Technologies | LSI Europe '23

Speakers

Henry Peck

Henry Peck

VP, Growth & Strategy, LSI
Read Biography
José Pedro Almeida

José Pedro Almeida

Chief AI Strategist (Various Companies), Advisory Board, Intelligence Ventures
Read Biography
Lina Williamson

Lina Williamson

Head of Entrepreneurship, Hospital Clinic de Barcelona
This panel aims to guide pre-commercial and commercial stage startups in successfully entering their first hospital and how to navigate the path to further expansion.

 

Transcription

Henry Peck  0:05  
Welcome to day three of LSI Europe 23. I'm thrilled to be joined by Jose Pedro and Dr. Lina Williamson. Let's start with some intros. Lina, what have you?

Lina Williamson  0:14  
Yeah. Hello, everybody. My name is Lina Williamson, the currently the head of entrepreneurship entrepreneurship for Hospital Clinic in Barcelona, also part of the MIT Sloan fellowship for catalysts. Previously, I was the director of the translational accelerator at the Brigham and Women's Hospital in Harvard Medical School in Boston. And I'm a serial entrepreneur, I've created about 16 startup companies, I own some of them. And before that, I was at Novartis for 10 years, also doing some science and business development, strategic alliances with academic institutions. So I've been in innovation and entrepreneurship, most of my career.

Henry Peck  0:58  
Great, thank you. And Jose Pedro. 

José Pedro Almeida  1:01  
Morning, everyone. I'm Chief AI Strategist for several companies nowadays. Also advisory board member for venture capital in the US, which is Intelligence Ventures focused on investing in LCI. My previous career, I work 10 years in the largest hospital group in Portugal, where we've lived, one of the most globally recognized Big Data and AI platforms. And afterwards, I was hired by the largest diagnostics group, here in Europe, which is called Union Labs. And where I've worked for, for the CEO in Portugal, leading all their data intelligence, transformation, which become became a case study for their MBAs.

Henry Peck  1:54  
Fantastic. And obviously, a lot of experience on this panel across large corporates, hospitals in different technology, verticals, diagnostics, AI, but let's focus on the hospitals for a moment, because in this audience, and at this event, we have a lot of medical technology, health technology companies that talk about how their revenue, their partners are going to come from these large hospitals. But I want to really understand from the inside out rather than the outside in what it's like to build a practice in one of these hospitals and get the adoption that they're talking about. So talk to me about the state of affairs at the hospitals that you've worked at when you started, and how you started to build that practice and secure that buy in to bring emerging technologies into the workflows. And we'll start with Lina.

Lina Williamson  2:39  
So in my current hospital right now, the Department of Entrepreneurship did not exist. So in I came into the hospital and build the the Department of entrepreneurship, and because of my experience, you know, looking for needs inside into bringing innovation, from the outside in, I started to support the whole hospital transformation, identifying innovation from the outside, and bring it inside, and also evaluating what was going on inside and what needs we would cover. So the organization from the inside, before I was there. It was the Department of Innovation was just a typical technology license office, if you will, but it was not set up for transformation. So now we are a little bit a bigger group, you know, we have identified the needs, we are in the process of identifying the needs, because we are a large hospital, we are an 800 bed hospitals. So I'm divided in different Institute's and the structure also kind of, it's complicated, because it's not a typical flat organization. So we are in that process, creating multidisciplinary teams and the whole decision making process so that we can better validate innovation, and implemented in a stream, you know, in an efficient way, instead of like one off here and there. And then let's see how we that's how it was before. Let's see what happens. Okay, you bring me an innovation and we figured it out. Now we have a process. So I think that will help.

José Pedro Almeida  4:18  
So on my case, same way as Lina, I've created almost 15 years ago, one of the first data and AI departments inside a large hospital, but when I got there, it was really hard. You know, almost 400 million euros per year annual spending 1000 beds, 6000 employees, 60 departments 100 different IT systems 30,000 people coming in and out every day, and I was the youngest director ever there. So I was working with you know, senior, very senior doctors, like 60 all see A few years old and onwards. And what I found out is that you really need to trust is something that takes time to accomplish. And so they need trust in you. And that process is something that you need to work together and be part of the team, be always available. But at the same time, I must say that if you are going for a transformation, like I've led, you need to have some empowerment. And I was speaking to some folks yesterday that are trying to sell to these large hospitals. And then for instance, they are blocked by the IT of the hospital. So one of the things I've learned throughout my career is like, in my case, because I lead the transformation from the inside, I always answer the CEO. And that makes a lot of difference. Because you know, it opens the doors. And the message is being is coming from the top down for the IT guy, you know, to to allow you to connect the cables, and, you know, show your stuff.

Henry Peck  6:07  
I want to follow up on that point about trust, you talked about how important it is. And there are a lot of, you know, gaps or riffs, maybe between building a new department like this inside a hospital, the types of individuals that may be on that team, and the other individuals, the other stakeholders in that hospital, the executives, the clinicians, how do you position some of these exciting but a little bit black boxy technologies to some of these clinicians, when we talk about AI that may automate certain parts of workflows they're very familiar with, or parts of procedures that they've been doing for decades? How do you come in as somebody with a different skill set and get that trust? And what type of people do you put on that team around you to support and reinforce the trust that you're trying to garner?

Lina Williamson  6:51  
I think the key to that is not to impose, you know, and work with the users. So my way of managing without power, because that's what it is, is to involve all the different users in the decision making process. So you know, it's it's like, in some decades ago, when we had electronic medical records that basically were imposed, you know, and the doctors were not asked if they were going to use it, because initially, it was for, you know, reimbursement. So you know, they kind of didn't know, because it was with administration, as they evolve, they started to be involved, but we don't want to do it in that way. So for me is really getting to know or getting to ask them what they want. So I usually go like, What is your wish? If we had all the resources? How do you envision your department, and they make them talk? And that's how I figured out the needs that they that they have in their departments? And then who actually uses that? Is that for the nurse? Is that for the patient? Is that for this dashboard? What do you need? So it's just interviewing, interviewing and extracting information, once I extract information, then I empower them to be part of it. So all I do is, for example, the last collaboration we did with Amazon Web Services to empower to support our workforce identifying, you know, digital solutions for them, it was exactly that it was like, Okay, what do you need, you know, and then who's going to use it, okay, call me evaluate these opportunities that I have, do you think that you will, that you will use it and then put them together in a room with the company and I am out in the company with the company with the team and the team is doctored nor is it can be a porter in there, it's like, everybody is not just the hierarchy, you know, is flat, everybody feels empowered. And then they can identify, you know, the company can really see how these people work. And the people who work inside the hospital, feel that we are listening. So and then as they share, then I said okay, how you're gonna pilot it, you know, how you're gonna do so I'm just guide them, but never ever impose. And I think that is the way of gaining their trust is when the final word is from the end user, of course, when it comes to purchase and everything goes up to direction and the direction will kind of help to decide, because the money is managed, you know, centrally. But, you know, at least the first stages of identification are totally empowered by the organization.

Speaker 2  9:33  
So, I couldn't agree more with Lena. So I back up everything that she said. I think that, you know, from a more agnostic point of view, doctors and nurses, they don't by the AI algorithm, or you know, any fancy slide, they buy, mostly impact. And for that, I think that are two keys. Just one is this one that Lina was telling about if you have the possibility to build together, so being on the ground with them every day understanding their most fundamental and clinical issues, that will be key to build the trust. Second thing, I think, you cannot ever sell the message like, you know, this product is going to diagnose better than you. That is absolutely the wrong message, for two reasons. First is, you know, you will find the name and enemy on the other side, which is something that you don't want to. And the second is that you are being quite naive. And the reason being is that, you know, I've worked with doctors for 10 years, every day on the grounds. And I learned to respect a lot, you know, that magic art that sometimes they have in diagnosing patients, and so don't sell the products in a naive way, like he's going to diagnose better. But to finish my point, I can tell a very short story that resonates with this. So basically, one of one of the systems were built at that hospital was, you know, sending alerts for patients at risk at scale every day. So sending alerts to mobile phone to the mobile of each physician. And it would, it could predict seven days early, earlier 30% of ICU admissions, and it was back then, and still today, one of the most advanced systems in the world, but the reason was only young doctors use it. And we tried everything to convince the senior ones, like put posters in the entire hospitals, send, send them emails, go talk to them, it didn't work, you know, what works, what worked was a Friday evening, a text message was sent to a general surgery department doctor that was in charge of 50 beds, signaling that there was a patient with a severe mo globin drop. And he was like, nobody told me this, he went to talk with the nursing team, the nursing team told him the patient was fine. It was like Okay, I will see the patient, you went to see the patient, talk to the patient, the patient was fine. He asked for another blood sampling. When the other blood sampling came along, the severe drop of haemoglobin was even higher patient was taken into the emergency room, and we saved his life. And so that precise moment, spread the word of mouth to the entire hospital. And suddenly we have all this, you know, senior doctors with us, supporting the system. So it takes time, but mostly takes the impact.

Henry Peck  12:48  
Yeah, proving that case within it's such a real world way that's central to what they care about, which again, is delivering the best care possible. It's not the slide deck, it's not the creative marketing, it's not the big pushes. It's validated technology that really makes an impact. And so having seen many technologies from the outside, in technologies that you've incubated inside the system inside out, talk about the process of implementing this. Now, inside you have your department you have your team, you have buy in for this practice. Give us an example of a technology maybe in the hospital in Barcelona that you've implemented, that was very successful, and then maybe some more perspective on where companies that are trying to get their technology implemented go wrong in this process of pushing for adoption in the hospital and failing to get the adoption that they need.

Lina Williamson  13:38  
Um, I think what comes to mind is probably the technology for surgery, Proximie. And they initially that technology was embedded into another big company technology. So physicians already kind of knew about it. So that was easier for me to bring it in. But then we as we brought the whole package of the technology. The first stage was, as I mentioned, you know, make them work together with the users but also bringing in informatics bringing in infrastructures, you know that you don't think about the engineers because the surgery room had to be changed, you know, you had to be put cameras and so not only having the doctor, the porter, the nurse, but also having the it having all the people who are going to be involved in the process, add ones in the same table, so they can talk about the different aspects. That was a key for us to first get in into a pilot because we have to pilot we have to pilot and I know we have a terminology that is pilot Tyrese and we don't want pilot IDs. You know, we don't want to pilot and pilot the pilot, but pilots something that we really know we are going to implement but the pilot is just to identify those little parts of the technology that need to be modified, so that it can be streamline implemented. So, once we have identified that, for example, in this case, we realize that we want, we had 32, surgery, surgical rooms that we wanted to transform, but we needed voice recognition. And at the end, they said, Okay, if we don't have voice recognition, we do not continue. So we could come back to the company and say, Okay, we need that part. And they that part was under development. So then that company could actually say, you know, what, let's demand and then develop that part, make sure so that, that we can implement the solution in the way you want it 100%. But they were really flexible in both sides, because there was a relationship already established, to say, Okay, we're gonna bring voice recognition, and then we're gonna go forward. So and I think that's the key, you know, the key is, people to people is relationships, we do business with people not with companies at the end. So, you know, make sure all the stakeholders are there, including the ones that you don't think about, like infrastructures that you usually call later, and then you realize, like, oh, you know, I can't modify that. And then working together as a true partnership, so that companies that don't have fully developed solutions can actually develop it with us. So I think that's the key for it, you have another example. So

José Pedro Almeida  16:31  
So I think the first thing about the team, which is important is complement the internal resources with external expertise. Outside In, I've done the same thing, and it was really valuable. But one key point was that I always told that external team was, you need to be inside. You don't, you are not building this from the outside, you need to be inside every day. Because that makes a real difference. Because you know, those doctors and nurses that you are going to work for at the end of the day, they will see you there. And just by seeing you there, it's different, you are already part of the team. And I advise any company that is trying to sell to, you know, one of these large organizations to be there, even if it's losing money from the beginning. Because it's like a family, you know, you start to be part of the family. So we I brought those, you know, external experts in, we build 560 billion data points platform that could basically answer any question that you can imagine inside the hospital. And when we started helping doctors and nurses with any question like, you know, how many patients felt yesterday in the third floor with more than 50 years old, that are diabetics and live outside the region like this. And they're like, Oh, this is interesting. And then you do this over and over again, and they started started coming to you. And they brought us a very serious issue when they started trusting in us, which were hospital acquired infections. And they were like, we need to isolate patients with these superbugs that are one of the worst, global health problems faster. And this to say that time in healthcare in clinical practice is really key. If you can reduce time to action, that's really key. So basically, we, we use that big data platform to alert doctors and nurses, when we found one of those superbugs by analyzing the lab results in an automated fashion, figuring out where were the where was the patient at that specific time, and sending the message to the nursing team to isolate that patient. So to give you an idea of the impact, this simple thing brought, two years later, that same hospital dropped 30% hospital acquired infections. What that means, at the end of the day, in that specific scenario was 15 million euros annual savings just just with this.

Henry Peck  19:12  
When you talk about the opportunity to bring some of these solutions in the opportunity to be part of the team, I think one of the things that, you know, having come from the innovator side and thinking about putting myself in the shoes of some of the executives listening, a lot of it might be that's great, but you got to get in the front door first. Right? And you guys have been there looking at solution providers in the outside. And what's the thing that hospitals always want to see is validation, right? You want to see that this works, you want to see that people have used it and done the kinds of things that you want to do. And so you end up in this chicken and egg problem is, as we've talked a little bit about earlier in the week, where you need the validation to be able to get it into the hospital, but it's getting into the hospital, it's going to give you this validation in this specific setting. So what's a startup company, an early stage company to do to get that validation or at least incentivize you to take that chance to be the one to bring that validation for them. 

Lina Williamson  20:05  
Yeah, I think that's a very important question. I am actually in both sides on the seller side with the startup companies that I that I have,

Henry Peck  20:13  
Say more about that, can you explain a little bit more about how you're on both sides of that equation?

Speaker 1  20:17  
So, you know, from the buyer side, or from the hospital side, usually we want, we ask the question, have you Are you already implemented in hospitals bigger than 500 beds, and those are kind of the first line innovation that we'll look at a say come in, you're already proven come in. But then when you are on the other side, and the setup side, and you have investors, you gotta find your first client, you gotta find your first client. And it's so hard to find the first client because that first client is asking you to have to be already validated. So I've been thinking about it. And I think the the solution, probably too, it is also from the investor standpoint is give the startup a little bit of a briefer terms that the startup can use to go into the first client. So the first time has some benefits, you know, I can open the doors to the startups and say, Okay, I can help you validate your innovation. But what isn't it for me, you know, can I have a, maybe a free use the first six months, then 50% card the first year, and then maybe I can end up paying full price after three years or something like that. Some sort of negotiations with the hospitals that I know, hospitals will be open to do, because investigators, at least in academic medical centers, like ours, everybody is open to, you know, help with validation and write papers, which is good for the company as well. So I think it would be important to have these preferred terms for the first client, which sometimes and I've seen it with my with the startups that I helped, we don't think about it. But maybe the board can also say, Okay, what about these preferred terms for the first time, just give us some incentives from the buyer side, to include this new technology to co-develop in order to publish the you know, the findings, and then work a little bit towards finally getting to purchase that innovation. And then that will open doors to many things. Because our hospital for example, we are a flagship hospital here in in mostly in Catalonia, but also in Spain. So this is a public system, right. So everything that we do, and we implement, the whole system will see it. So if we have validated with one of the startups, even if we have preferred pricing, then it's very easy to spread and to gain a second or third and a fourth client. So just think on that, on that process. From the startup point from the investor side, don't be too pushy. I mean, of course, be pushy on the first client, but help a little bit your your startups to get there.

Henry Peck  22:57  
I know you have experience on both sides, you mentioned that in a public system that information can disseminate more rapidly maybe, does this game change when it's a private system that we're working with? What would you do differently?

Lina Williamson  23:09  
Well, you know, with the public system, the problem is, is very slow, very, very slow. And the purchasing process is very bureaucratic. So maybe in the private system, you know, is faster and is easier, because the decision making of purchasing can be more streamlined. So you either go into the pot into the private system, and you work with one hospital, well, if you're with a hospital with a partner system like mgv in Boston, then you probably can have access to seven hospitals or something like that. But most often, you know, you have only one institution. So you either do a good deal with one institution and is faster. But going to the public system, it can be slower, but then you can actually spread a little bit further. That's how I see it. 

Henry Peck  23:56  
Make sense? My experience, appreciate it. And I know we could talk to you about your experience in the hospitals for hours. But one thread I want to pick up on Jose Pedro with the last few minutes here is what we're going to be implementing in the future. And you were implementing Big Data AI before it was the cool thing of the year, years ago. I want to talk about what you're seeing in the AI generative AI space. And when you're looking at the health technology, opportunity broadly, what should companies be thinking about? What should they know? With your AI expertise, how can they position themselves to have a real impact in the hospital systems of today and tomorrow?

José Pedro Almeida  24:37  
I think we are in unprecedented moment. In AI. It's an inflection point. Because the way the whole way, everywhere everybody was doing AI, myself included for 10 15 years. Just you know came to an end and what I think is that startups and companies that are relying their business models on this old, traditional way of building algorithms, and selling that are at an existential risk at this time. And the reason being is that these large language models that most consumers see as ChatGPT are the most powerful technology I've seen. It evolves at an exponential rate. So you know, in six months, it has evolved more than any other technology in 12 years. And what that means as well is that if if you have a company that's relying on a certain algorithm that you know, is, for instance, predicting lung nodule in a CT scan, and that's their business model. The thing is that these logic language models are multimodal. So they will be able not only to see that image, but correlate that image with the clinical notes with the vital signs with the lab results with the pathology reports. And that is a huge advantage when compared with your product. Because the tech the startups and companies that start building that their tech stack on top of this first, they will be much they will have much more power in terms of the way they can predict events. And secondly, they will be able to absorb a lot of more data, more data points, different data points that are really key for the doctor's decision making process because the doctor does not decide if a patient has certain tumor or not just based on a single point, it decides based on several. So I think that we really need to look at what this technology is able to do. And just to give you an example, we are seeing that, for instance, these lead language models, and Google is doing a lot of great work there can find tuberculosis in an x-ray, without ever being trained in X-rays or tuberculosis. Traditional process like you show them an x ray and the model says tuberculosis is there any any explains why.

Henry Peck  27:20  
Explain why that's possible. Now, though, because like you're saying that traditional way would be, you know, you show it a bunch of pictures of a dog it learns what a dog looks like. And now it knows that this is a dog it classifies it. Yes, what's happening now, like you're saying is this is able to go well beyond that. And so if you have one of those traditional models, how do you future proof your business against what's what's happening now? 

José Pedro Almeida  27:40  
You need to you need to pivot fast, because, no really, if I was investing, or if I was in one of those companies, you need to pivot fast, because this is taking over this is you cannot compete with something that you know, first, as learned with the entire knowledge of the world. And when applied to healthcare has been fine to tune to healthcare. That's true. But the traditional way of doing things is coming to an end, I was talking to a company in Boston, whose business model is classifying medical images. So they crowdsource they pay doctors all around the world to classify medical images. And it has been their business model for years. Now the thing is, these models now can automatically classify their own images by reading, you know, image reports. They don't need a doctor anymore to do that dog, not dog thing. So their business model is at risk. 

Henry Peck  28:47  
Makes sense. And for our last moment here and Lina I'll pass it to you as we have Jose thinking very critically, and both optimistically and a bit, you know carefully about the future of AI in your hospital right now. What would be your perspective in response to that as you're evaluating the landscape of large language models and generative AI? How are the leaders in your system thinking about it and any closing thoughts on how hospital leaders may be looking at the future of technologies like aI

Lina Williamson  29:13  
it's a little bit intimidating from the inside because we need to do a lot of education on the concept of these language models on the concept of how actually you're going to implement it and it's going so fast and I think for us implementing it it's a little bit too slow inside and I think that's a big challenge that we are having right now is like okay, I we need to pick a pace you know, we need to do it. But inside we still have scheduling in Excel sheets, you know, is like how am I going to go we can't I mean, we do the most humane way possible to go as fast but it's really tough. So I think that we need to identify ways to educate fast our our our people in prepared I mean, in small teams and do small implemented implementations that can then spread around transversally. But that is a difficult task that we have at hand now. So we'll see where they're good.

Henry Peck  30:13  
Well, we will see for sure. And thank you, everybody for joining us. That'll conclude this morning's keynote. Looking forward to a great day three of you have questions for these to be coming off stage now. See you for the rest of the day of the panels of the presentations, partnering meetings and tonight's fireside chat. Thank you

LSI USA '24 is filling fast. Secure your spot today to join Medtech and Healthtech leaders in Dana Point.

JOIN US TODAY

Share this video

Companies We Work With