Amanda Herbrand, a clinical data specialist and doctor of oncology, and Bram Stieltjes, the Department Head of Research and Analytic Services at Universitätsspital Basel (USB) are spearheading a transformative journey towards the establishment of a data-driven hospital, leveraging the power of openEHR in the process. Their approach – blending clinical expertise with technological advancements – promises to reshape traditional healthcare models and drive meaningful improvements in patient care and decision-making processes.
AH: I’m a former oncologist, but now I work as a clinical data specialist and I dedicate my time to building our data driven hospital in Basel.
BS: Like Amanda, I’m also a doctor, but my specialty is radiology. At the moment, I lead the Research and Development department within our central IT. Our main focus is on finding solutions to issues in the hospital setting, particularly those where existing products may not fully address the need or where there’s a gap in the market.
Our goal is to stay at the forefront of innovation, but not just for the sake of it: we’re not interested in what I would call “sport shoe innovation” – all style, no substance. But rather, we’re dedicated to implementing meaningful changes in clinical practice.
We’re a diverse team of around 25 individuals, including doctors, full-stack developers, and the occasional physicist. Together, we tackle a wide range of challenges, all with the common aim of improving decision-making processes within our hospital.
How much of your time is dedicated to this work? Do you still work in your roles as doctors?
AH: I haven’t worked as a medical doctor for about two years now.. Beforehand, I dedicated a whole year to it, splitting my time right down the middle – half the week here, half the week there. But after that year, I realised it wasn’t something I could see myself doing for another year.
Switching to what I’m doing now was definitely the right move for me. It’s more enjoyable and keeps me interested.
BS: I can’t say I miss the patient interactions – I’ve been out of that environment for quite some time now. I spent years in academia: patient interaction wasn’t part of the picture, and it’s been the same for the past decade in my current role. It’s not that I dislike it, it’s just that what I’m doing now requires a lot of focus and dedication.
While I may not personally miss it, I do recognise the importance of having clinicians on our team who are actively engaged in patient care. That’s why we’ve made it a priority to support clinical roles financially. It’s vital they have the time and resources they need to contribute effectively. So, yes, we’re investing in those positions to ensure we have the expertise we need. And it’s not just about buying their time; it’s about valuing their input and expertise as integral parts of our team.
You mentioned 25 members in the team. How do your various roles break down?
AH: Implementation is just part of it: we’re also responsible for managing the data warehouse of the hospitals, along with handling data deliveries. We also conduct research and development in various fields. It’s not just about openEHR; we’re also working on building a genomics pipeline. The goal remains the same: ensuring data transparency.
BS: We’re able to create environments where we can extract clinical data from raw data and integrate it into clinical workflows. Currently, there are around 25 people working on this, but the team implementing openEHR consists of roughly three individuals. We’re still in the early stages there.
But to really scale this up, we’ll need a larger team.
How large is the hospital?
AH: USB has around 7,000 employees and roughly 700 to 800 beds. In other countries, the ratio of personnel to beds is usually lower. But Switzerland tends to have a higher density of personnel per bed.- it’s not just us; other Swiss university hospitals have a similar number of personnel.
Although our hospital might not have a large number of beds compared to some, we are still the biggest hospital in northwest Switzerland. The largest Swiss University Hospital has about double the number of beds compared to ours but even then, in international terms, USB is still considered a mid-sized hospital.
Can you explain the concept of a data-driven hospital and how it differs from a traditional one?
AH: Well, a traditional hospital is a bunch of neatly contained, separate sections. You have your different departments within the hospital with your separate processes and your various applications being used. But the upshot is – in a traditional setup – everything is pretty much closed off from everything else: each section is in its own little bubble.
Now, the concept of a data-driven hospital is to shift the focus away from just the departments and applications, and instead, focus on the data. Imagine the data flowing seamlessly from one department to another, all in sync with the patient’s journey, so essentially, we’re putting the patient right at the heart of our thinking, our designs and our processes.
BS: We need to take all the information swirling around the patient and structure it as data. So, step one is figuring out what should be at the core of it all. And then, the real magic happens when we start using this data to drive smarter decisions. Maybe one of the tangible goals of a data driven hospital is that you can create a self-learning system where healthcare personnel can learn from decisions made with similar patients and implement those in future decision making.
Sounds good, but what challenges do you foresee in implementing your data-driven approach?
AH: First off, there’s the status quo, which I’d say is our biggest hurdle in many ways. On the one hand, there’s the status quo of our applications and infrastructure, which is straightforward enough. Then there’s the status quo of people’s mindset, which is… a whole other ball game.
BS: Convincing the board was not easy. We did manage to get them on board eventually, but it was a real uphill battle.
We had to pull out all the stops, even flying out to Ljubljana to film evidence of a similar setup in a hospital. I mean, I found documentation, but that wasn’t enough to convince everyone. Criticism came thick and fast, even after the decision had been made over a year ago.
And I think without a proof of concept, we could never have convinced them. Even then, some people couldn’t wrap their heads around the idea. It felt like they thought we were making it all up. Seriously, it was tough going.
You’re talking about the openEHR proof of concept…?
AH: That’s right. To demonstrate the feasibility of a data-driven hospital, our primary concern was how to effectively preserve and store data for the long term, how to ensure accessibility throughout a patient’s lifetime.
We researched various options and openEHR emerged as the standout choice for us. Despite the time constraints we faced, it was the solution we needed to explore further.
Choosing openEHR for our proof of concept was a decision that paid off. Its storage capabilities and robust data definition were particularly impressive. The availability of a comprehensive data dictionary, rather than just a data index, was a significant factor in our selection.
BS: An issue we encountered with other systems was the lack of a consistent data dictionary, which meant that defining and accessing data across different trials or departments was cumbersome and often impossible. That disjointed approach to data management highlighted the importance of a stable data model, especially in operational settings. Persisting data but also describing it in a consistent and reliable manner was a priority.
What types of data do you intend to collect and analyse?
BS: We’re talking about data here, and when I say data, I mean all of it. Every last bit. Believe it or not, there’s no such thing as unimportant data. The way I see it, we’re keen on diving into every aspect of our processes, even the little side ones. There’s hardly anything we wouldn’t want to analyse. Obviously, when it comes to what data we want to make public, that’s where we draw the line.
openEHR and structured data is quite pertinent to us, in two main ways. Firstly, there’s the importance of high-quality data for training algorithms; without good data, algorithms simply can’t perform effectively. So, it’s logical that we’re using open sources to access the quality data needed, particularly for applications like clinical decision support in AI.
And on the other side of the argument, there’s the recognition that AI can’t magically solve all data quality issues. While it might help streamline processes or even assist in improving data quality, it’s not a one-stop solution. It’s more like a helpful assistant than a miracle worker.
AH: Raw data, like images, for instance, we’d rather keep behind closed doors. We prefer to share refined extracts instead. But anything we need for making clinical decisions or supporting our secondary processes, that’s the stuff we want in our repository.
BS: Yes, that’s our approach. There’s no data too small, just stuff we don’t need. And once we figure out what’s unnecessary, we can streamline our documentation processes. If nobody’s using it or asking for it, there’s no point in recording it, simple as that.
What are your long-term goals for the development of the data-driven hospital?
AH: First, establishing a data-driven hospital is a substantial, long-term goal for us. It’s not something we can achieve overnight but rather a vision we’re working towards, step by step. With that said, it’s crucial to understand that our vision extends beyond just one hospital. We’re aiming for a data-driven healthcare ecosystem, collaborating with clinics, hospitals, primary care providers, nursing homes and mobile nursing services. It’s about creating a flow of information throughout the entire healthcare network.
BS: We’re actively engaging with stakeholders, including our Ministry of Health and pharmaceutical companies based in Basel, who share our enthusiasm for leveraging real-world data. Moving away from traditional trial methods towards real-time observation is a significant shift that benefits everyone involved. By harnessing existing data sources, we can avoid unnecessary costs while still gaining valuable insights into patient care.
Collaboration is the key. We’re in talks with other hospitals and GP groups because we understand that patient data shouldn’t be confined to one institution. It’s about ensuring that data follows the patient wherever they go. Otherwise, we’d be stuck with outdated practices like receiving faxes and manually entering data, which would be both inefficient and prone to errors.
AH: Our efforts extend beyond our immediate region. We want to inspire similar transformations in other healthcare settings, by leading by example, The Karolinska model, which revolutionised healthcare in its region is a good role model of how a hospital can change the healthcare environment. So, while rolling it out on a regional or national scale – like in Catalonia and Slovenia – may be challenging, we’re committed to driving change like Karolinska.
Establishing a data-driven hospital is our primary goal, but our broader vision is to transform the entire healthcare landscape for the better. openEHR will play a pivotal role in that. Rather than relegating it to research or analytics, we really want to use it on a real-time basis to feed the applications we’re using in the clinics, and enhance patient care.
Leave a Reply