Insights based on data can lead to more efficient operations and greater care quality. However, a great deal of the data available remains unused. Big-data specialist LynxCare wants this to change – sooner rather than later. LynxCare has built an AI platform with the goal of generating all kinds of relevant insights that can improve patient care and stimulate research. But to power this platform, you need data… lots of data.
To unlock hospitals’ data securely and efficiently, LynxCare relies on Amaron. We asked Georges de Feu, CEO of LynxCare, and Maurice van Kerkhoven, co-CEO of Amaron, how they accomplish this.
> Georges, can you describe in a few words what LynxCare does?
We are focussed on everything related to data in healthcare, and to its accessibility. Everyone knows that the healthcare sector has a massive amount of data. But it isn’t being used to its fullest yet. This means that you, as a patient, must sometimes tell the same story over and over. Alternatively, as a care provider, you want access to insights from the existing data, but these insights are not available. Five years ago, we were thinking about solutions to this problem, and that led to the foundation of LynxCare.
However, making data in healthcare accessible is not straightforward: information is spread out over many different data sources, and is often not structured.
One of the first major challenges was to find a way to make this unstructured data accessible. We quickly noticed that there wasn’t technology readily available on the market that could handle this adequately. That’s why we built a big data platform ourselves, based on AI and NLP.
Two years ago, within the context of a European data network, we launched several broad research projects related to COVID-19. We then made large quantities of non-structured data from the medical records of different hospitals accessible. For LynxCare, this has been a catalyst for data mining and for structuring the care data for research purposes.
As a result of the pandemic, many more people are now aware of the importance of high-quality data, and we are seeing a huge acceleration in the field.
> Who are LynxCare’s customers, and in which care domains are you active?
GDF: We work for hospitals, helping them design robust clinical databases or data warehouses for research. The hospitals themselves are responsible for involving the appropriate research partners.
Apart from COVID-19, there are, of course, a whole range of disease domains where there is an enormous need for data-based insight. For example, cardiological and oncological disorders, which have a much higher mortality rate than the coronavirus. For these types of pathologies, it can be very useful to assess as accurately as possible the effectiveness of certain interventions or treatments; data analysis plays an important role here. We currently have projects of this kind running in several hospitals, including AZ Groeninge, AZ Monica, Ziekenhuis Geel and Maria Middelares Gent.
At University Hospitals Leuven, we are working on a project related to rare diseases as well. Making these diagnoses for patients can be like finding a needle in a haystack. Technology, combined with a lot of data, can certainly help medical specialists to detect rare diseases more quickly.
Our work now increasingly reaches beyond the Belgian borders, and we aim to grow internationally. We already have customers in the Netherlands using our platform, and we intend to expand our activities to other countries as well, soon.
> To achieve results, you need a lot of data. How do you approach this?
GDF: Hospitals don’t have monolithic IT architectures, but rather diverse landscapes with a myriad of IT systems. And within these systems, there are many types of data: both structured and unstructured. We let our NLP technology loose on this data with the aim of improving patient care and stimulating research. That’s our core.
We want to continuously mine the sources in the hospital, and make the right data available.
Of course, connectivity plays an important role here. But it isn’t our own core expertise, so we collaborate with partners who know the various subsystems used in the care sector through and through, which enables them to create and maintain effective interfaces.
Amaron has this expertise, plus it is a highly trusted company in the sector – which is also an important consideration. That’s why Amaron is our preferred partner for connectivity.
> Maurice, what specific role does Amaron play?
We form an important link, both on the LynxCare side and on the hospital side. There are gateways on the two sides, which ensure that data transfers occur in a secure and efficient manner.
Together with the hospital, we determine which data set to make available to LynxCare. We believe access should only be granted to the data that is strictly necessary for the project. We can also anonymise data, when necessary.
LynxCare then receives exactly the data they need—easily and in the correct formats—ready to be analysed and processed.
> How did this solution come about?
MVK: Through healthy interaction between LynxCare and Amaron. We searched together for answers to a whole series of questions. What exact data does LynxCare need? What is possible on the hospital side? How can you implement this? Is it realistic? How do you ensure that everything is secure? We each came to the table with our needs and arguments, and there have been several iterations. Ultimately, we created an architecture that is widely supported and takes into account all of these aspects.
GDF: In addition, based on pilots and hospital feedback, we have further optimised our setup.
The result is a standardised product that we can roll out to any hospital, and which we can propose in partnership to the entire sector.
> How do you guarantee that medical data is handled securely, and that patient privacy is protected?
GDF: When a hospital decides to take part in a research study, we always consider, together with them, whether everything is in line with the patients’ expectations. The hospitals decide for themselves for what purposes they will use the data, and they always retain control over their own data—it is and remains theirs. Neither LynxCare, Amaron, nor any other party can do anything with the data without the consent of the hospital.
Besides privacy, there is, of course, also the security aspect. We work with medical data, which means you must always adhere to the highest standards. Amaron ensures that data exchanges are carried out according to the latest standards. You know exactly what data goes where, and everything is secured end-to-end.
In addition, we regularly engage ethical hackers to conduct tests of our infrastructure, to make sure that the installation is highly robust and that there is no possibility of data leaks at any time.
MVK: The communication between Amaron and LynxCare is hyper-secure. We test this communication, and the technology we use has definitely proven itself in terms of security.
Furthermore, we work within the hospital’s own IT environment. All data sets that may be used for analysis are filtered: which data gets through is determined at the hospital’s port. Plus, we ensure that the data is anonymised or pseudonymised, if necessary, to further reduce risks. We make specific agreements related to this with the hospital for each use case.
We never store data from the hospitals on the Amaron infrastructure. We have secure connections that enable us to access the hospital servers or data via VPN, if necessary. We always do this in consultation with the hospital. Each transaction, each message that is sent is logged within the connectivity environment.
Accreditation helps us to further ensure our comprehensive approach. A few years ago, we received ISO 9001 accreditation, and we are currently working on ISO 27001, which focuses on security and the GDPR.
GDF: At LynxCare, we are also ISO 27001 compliant, and we conform to the GDPR regulations. Because we are active in the Netherlands, we also abide by NEN 7510, the Dutch standard for information security in the healthcare sector, which is even stricter than ISO 27001. If something is not sufficiently secure, it is a no-go for all parties involved.
MVK: Apart from the strictly technical aspects, there is also the fact that three different parties are involved in each project:
- the hospital itself, which is careful about making data available;
- Amaron, which has the data-level expertise, techniques and processes;
- LynxCare, which puts everything in place to ensure secure data handling.
As a result, there are three layers of security, each implemented by a party that is highly focused on security, and that fully understands the requirements when working with medical data.
> LynxCare’s data analysis can also add value to Amaron’s products. What opportunities does Amaron see for working more closely with LynxCare?
MVK: Our current projects, including in cardiology and orthopaedics, were started on the initiative of the hospitals. But we have access to much more data than what is being used at a certain point for a specific study. Our Workflower applications, for example, also generate interesting data: from the processes we have set up for infection prevention, to give one example. We could analyse some of these data sets. Perhaps there is data missing that we could then generate. Or it may be useful to return certain data to the source system to arrive at new insights. There are a whole range of use cases I can think of in both directions; there certainly are opportunities here.
GDF: I agree with that. What we do is to build up a clinical data warehouse and try to increase as much as possible the quality of the data that we are making accessible from the source systems. But a data warehouse is no more than a database, and it has to prove its worth in practice. In addition to population insights and pure research projects, there are many possibilities in patient care where Workflower can play a role.
> How do you see future evolutions in the use of (big) data and AI applications?
GDF: When you talk to patients, they often assume that data in hospitals today is already accessible, analysable, and useful. When they see the actual state of the data, many are totally surprised. There is still a lot of work to be done …
Today, a clinical warehouse with NLP/AI is something of a luxury for a hospital, generally reserved for larger facilities. But I expect that in the future these data warehouses will become essential for every hospital.
After all, healthcare is increasingly based on data. So, you must make sure that all data sources are accessible and used to the fullest.
With the evolution toward value-based healthcare, hospitals will have to demonstrate that the care they provide leads to effective outcomes. And, for that, you need data.
MVK: I share Georges’ view. There is a lot of data in hospitals, but simply having it available is not enough. Doctors only have limited time with each patient. If they can gain insights based on data analysis, they can help their patients faster. And the need for this will only increase. More and more useful data is being generated; this data can help in diagnosis. The entire preventive aspect of care, for example, will be increasingly based on data.
At Amaron, our core task continues to be to connect all the data points together. Today, this happens primarily at the hospital level. However, with the current level of networking in the sector, it will become necessary to combine data from multiple hospitals and data sources.
Amaron will remain the player who knows where the data resides and how to make it accessible. We can combine this data exchange with Workflower to implement the right processes and to allow data to flow back, or to collect or create additional data for additional insights.
Has this interview with Georges and Maurice inspired you? Would you also like to make data securely accessible to external parties, but you aren’t sure how to do this? Then please don’t hesitate to contact us.