By JULIAN BEIMES
Hospitals, emergency and operating rooms are crucial components of our healthcare infrastructure. For years, medical doctors have worked in the same environment. With the rise of technology, one of the oldest professions on earth is now bound to change. Efficiency potential in healthcare appears almost limitless, but visionary talks are often detached from reality. We had the opportunity to talk with Gerd Schneider and Raluca Pahontu, who have worked on a fully connected operating room at the university hospital of Heidelberg/Germany (UKHD). In this extended interview, you’ll read about challenges and opportunities of a connected operating room, drivers of this development and an outlook on the future.
idalab: For our readers, who have not been to an OR for a while, can you describe how data and technology are already now a part of the OR?
Schneider: During active surgeries, doctors have access to a couple of systems, where they can access the relevant information, say laboratory data and radiological images. There are various displays and doctors would literally consult all of them to arrive at a valid overview. At the same time, there is an extensive documentation system, which gives an accurate protocol of OR procedures. Given this setting, the OR team really has to be on top of the game and needs to know precisely, on which monitor to look for which information.
What data of medical devices is already being utilized?
Pahontu: One of the main challenges regards the fact that we have devices and medical technology of various different manufacturers and most of those devices simply cannot communicate with each other. There are some manufacturers, which offer fully-equipped proprietary ORs, but in that case hospitals are dealing with monolithic systems, which has its own downsides. With the publically funded project OR.NET, integration and interconnectivity of medical devices with each other and with IT systems has now been significantly improved. This was also a major enabler for our project, as we were able to integrate open high frequency data, which we had never done before.
For example, we can now acquire data not only from patient monitoring and anesthesia devices (like heart rate, temperature, etc.), but also data from surgical devices (like light intensity from an endoscopy camera or insufflation pressure from an insufflator) in a standardized manner.
What eventually sparked the project at UKHD, why did you decide to work towards a fully connected OR?
Schneider: Initially, we actually didn’t have the fully connected OR in mind, that was more of an attractive use case. The project came into existence as we had a highly heterogeneous landscape of data sources at UKHD in general. There have always been requests to somehow work with the data, but our response was constrained by the fragmentation of data. So, we started off with the goal to establish a coherent data platform and architecture, which we could then gradually develop further. This data platform is not only crucial for clinical care, but could also be utilized for medical research. At the same time, having this data platform now allows us to explore analytics and data science together with the different departments
How did you work together in the process of defining first use cases?
Schneider: The surgeons at UKHD are actually very active with high IT affinity, they are definitely interested in exploring all possibilities of technology. Since we do have an IT background, they provided us with the first set of use cases.
Surgical assistance systems, for example, are of course one key use case field. But before I can even think about those kind of decision aid systems, I would need to acquire additional data, which comes a lot from process optimization. “How long is this surgery still going to last?” That sounds like a trivial question, but is extremely complex, if you break it down as it involves a lot of contextual information. What is actually happening in the OR? How can I draw connections from device activity to surgery duration? How can I reliably predict? That topic is of high relevance for us here, as it essentially has a large influence on surgery planning and optimization. The OR is the most expensive item in a hospital, thus capacity utilization is key.
Another area of application is real-time patient routing in a hospital context. If you know precisely about waiting times at different hospital wards, you can route the patient accordingly. So, instead of getting the ECG, the patient might go to get blood tested first. While we are still far away from such a scenario, it shows that there are plenty of use cases available, which go beyond the connected OR.
When having agreed on use cases, how did you proceed? What challenges arose?
Schneider: We do have a lot of powerful technology in the background, but if we don’t have the relevant data point available, it doesn’t really help our endeavors. We first assess our existing data sources and check, whether whatever is available is sufficient for the use case. If not, then we see if the medical device manufacturers do have data, which could help us. But oftentimes we also discover that we are simply missing the appropriate data, to fully achieve our goals. Let’s take the fully automated OR planning for example. To do that in real-time we would need schedules of resource availability, cleaning service bookings and other information. It is only in the process that you discover all kinds of data, which you are currently not maintaining digitally, but which would be extremely valuable. So we always keep an eye on: What is possible? And what is feasible? Because, getting a highly calibrated sensor system for our hospital would be possible, but it just not feasible at all.
Are data integration, data science and analytics a strategic focus of UKHD?
Schneider: We are still at the very beginning, but I think the mindset that hospitals, which analyze their data to improve processes, have a significant competitive advantage, is slowly arriving on executive level. We are oftentimes working with third party funding, which allows us to push certain aspects in a more targeted manner. But the focus of the hospital will always remain on clinical care. So, we can’t really say that we are strategically building up this great hadoop-cluster, which allows us to do fancy analytics. But we do appreciate the great potential, which a more enhanced utilization of data sources could bring in terms of efficiency for UKHD.
What’s the current state of the connected OR at UKHD?
Schneider: We do not have an infrastructure rolled out in the entire hospital. Instead, we are currently working with an experimental OR at the surgical ward. But that infrastructure is up and running and data flows have been established. So we are currently in the phase of trying to do first analytics and have developed some basic algorithms. Our main task within the last months has been to merge and integrate all different data sources, which we have accomplished. In the data science and analytics phase, we are now cooperating with institutions and universities to fully tap into the data potential.
Do you think hospitals will continue to be the driver of integration and inter-connectivity or will medical device manufacturers take over?
Pahontu: For the operator it is always the most interesting question to ask: why would it be beneficial to achieve inter-connectivity? If you don’t have a monolithic system, but an open and more connected system, it is definitely easier to replace devices in case of malfunction, but this in turn then reduces the dependency on certain brands, which might not be a business goal of the manufacturer.
Schneider: Manufacturers are generally supportive towards more inter-connectivity, but for them it also brings a whole lot of further implications. Once they do have interfaces with other devices, they run the risk of dysfunctionality. To ensure high standards in patient security, there is still a lot of open questions regarding the implementation of a standardized communication protocol. While it opens up new avenues of innovation, for example semi-automatic steering in the OR and robotics, security will always be a prime concern. Therefore, I am really looking forward to see how the results of OR.NET are disseminated throughout the industry.
So, what impact would a fully connected OR have on product certification?
Schneider: The question of responsibility in case of systematic failure is key in this area. Who is responsible if something goes wrong? In a fully connected OR, it is not that easy to completely trace down responsibility and the source of the fault. That again does have implications on the approval process by regulatory bodies. While the currently applicable laws do specify the standards for approval, it remains unclear how this can be merged with the idea of inter-connectivity. If we, on the other hand, try to connect devices ourselves, we do not qualify as operators anymore, but become “manufacturers” in a broad sense, that is certainly also not feasible. Thus, there is a lot of uncertainty at the moment because these questions have not yet been fully resolved.
In all these settings, how do you handle data protection and security?
Schneider: Data protection is our utmost priorities. We don’t do anything, before we have confirmed its legality and ethical adequacy. Sometimes, we even play the issues to a dedicated commission for an ethical vote. Thus, the push towards analytics and big data will most probably remain centered on the hospital itself, as we cannot really outsource any data work, as anonymisation and pseudonymisation could not be reliably conducted, given the large amount of unstructured data involved. Thus, it can sometimes be a lengthy process to get things on the way. Whenever we work with external institution, we most often have to invite them over, as we cannot really have the data leave our ecosystem.
In light of your experience at the current project, how do you think progress in the field will evolve in the next years?
Schneider: We are quite excited about the road ahead, but it is difficult to give an exact prediction. There is generally some dynamic going on and the OR.NET project has also been very well received by device manufacturers and research partners. But it still remains vague how they will eventually proceed.
Pahontu: It will definitely depend upon how fast the agreed upon standard will be implemented. But until such a standard has been adopted and is actively pursued by all relevant parties, it usually takes some time. Additionally, hospitals can of course contribute to an acceleration of the process, if they incorporate the relevant requirements regarding certain supported standards when sourcing medical devices. But my guess would also be that it will still take sometime until the fully connected OR becomes widely used.