Johns Hopkins tech innovator: Breaking the chains on data

Heading up the Johns Hopkins Medicine Technology Innovation Center, Gorkem Sevinc, MSE, CIIP, focuses the bulk of his attention on partnering with clinical and research personnel within Hopkins Medicine to collaboratively build innovative Health IT tools. Software development, IT infrastructure, workflow and collaboration tools are all part of his purview.

Sevinc started the informatics research laboratory in the Department of Radiology at Johns Hopkins Medicine along with his director and chair-elect of SIIM, Paul Nagy, PhD. They grew it to serve all of Johns Hopkins Medicine’s Health IT needs, but the department of radiology continues to be the lab’s anchor.

HealthImaging/Viztek sought him out after hearing him speak at the 2015 annual meeting of the Society for Imaging Informatics in Medicine (SIIM). There, he stumped his session’s attendees by asking what they thought was the most important thing to remember about analytics. Data, outcomes, and visualization, they guessed. Wrong, wrong and wrong, responded Sevinc, who is also CTO of two startups, a medical imaging second-opinion service he co-founded and a mobile health company that supplies a platform for remote patient management. “The most critical part of analytics,” he said, is “the story.”

Sevinc spoke by phone from Baltimore, where the imaging departments of Hopkins’s two main hospitals and two imaging sites generate in the neighborhood of 700,000 diagnostic images per year. Here are excerpts from the conversation.

Q: If you had the chance to design the optimal PACS from scratch, what would it look like?

Sevinc: Enterprise image viewers, which is a web-based technology, with server-side rendering and zero footprint viewers—that is the future. I agree with my friend Don Dennison who predicts PACS will be dead by 2018, replaced by web-based image viewers providing PACS-level functionality.

If I were designing the optimum PACS, I would challenge why the RIS and PACS are still separate and different systems. With the advancements of DICOMweb and HL-7 FHIR, we’re going to have a lot of flexibility compared with the standards we used in the past. We’ll innovate a lot faster.

Another friend and SIIM Hackathon Committee Member, Chris Hafey, designed an open source viewer called Cornerstone, which utilizes DICOMweb. It’s amazing that anyone can image-enable their application by grabbing Cornerstone online—this is a prime example of open source and web-based technologies increasing the pace of innovation. Our goal should be to continue building and utilizing modern building blocks, which also help us concentrate on solving clinical problems. 

Q: What are the real problems?

Sevinc: In the HL-7 world, we are still struggling with doing integration. In the DICOM world, more or less the same thing. Hopkins is going through Epic rollout for inpatient clinics. We have been spending ridiculous amounts of time just doing HL-7 integration testing. In my dead honest opinion that is a complete waste of time. I do not want engineers spending time integrating one system with another. That is not exciting, that is not interesting, that is not what we should be working on in 2015. Hopefully, web-based technology is going to reduce the amount of time we spend integrating and increase the amount of time we spend innovating.

Q: How can new and emerging IT technologies and platforms make a difference?

Sevinc: The DICOM and HL-7 technologies alone are going to make a huge difference.

Some of the EMR vendors are starting to make their data more public. One of the biggest challenges in a hospital system is getting access to data. Yes, we have to protect access to the data, and we have to be very careful about who gets what access to which data, especially with HIPAA’s need-to-know requirements. At the same time, we need to think about how we can keep up with the evolving needs of care delivery.

We have patients who need more access to data. We have patients who want to see their imaging. In this day and age, we still send people outside the hospitals with CDs of their images in their hand. I’m sorry, but that’s crazy! I don’t even have a computer at my house that I can put a CD in.

Image-sharing vendors are making a good dent in this problem, and the RSNA Image Share Network is driving quite a bit of the change along with health systems. I believe we have to step back and think about the end goal: We need to give patients access to their data. That has to be the top priority. A patient that can access any movie anytime from their mobile phone, or can access their home remotely to check on how their dog is doing, will not and should not accept being sent home with a CD of their MRI.

To enable this access, we must create easy-to-use, web-based systems, and use that data to provide better ways of engaging patients with their care—not just getting access to their data but understanding their data. By data, I do not mean only images but the full spectrum of the EMR. We need to get to where they can connect with their physicians and ask questions about their imaging studies so they can understand what the images mean.

This is where we are going to have to go, and it’s going to happen as we are moving from volume-based care to value-based care. We have to do these things—no question about it.

Q: How has the emergence of the EMR impacted what PACS needs to do?

Sevinc: We are right smack in the middle with this. With the EMR, people need access to their imaging. Traditionally what we have done at Hopkins is, we would give full-blown PACS access to anyone clinical who needed access to images. To image-enable our EMR, we rolled out an enterprise image viewer that is zero footprint three years ago.

The full-blown PACS is not what most clinical personnel need. Such a system is hard to manage on 55,000 workstations when you are trying to do routine software updates. Using a web-based, zero-footprint viewer for clinical personnel is a no-brainer for many reasons, among them ease of server-side management and having one link within our EMR that can be clicked to launch images directly without having to search for them.

Q: What is the ideal PACS-EMR relationship?

Sevinc: You have to take a step back and think about what the role of PACS is now and what it should be in the future as the EMR is starting to take on the RIS role as well. As part of our Epic roll out, we are deploying Epic’s RIS module, Radiant, which is essentially a RIS deeply embedded within the EMR. Subsequently, as the EMR takes on the RIS role, it’ll need to be context-integrated with the PACS to launch images and drive the radiologist desktop workflow.

Of course we also have and need the HL-7 integration between the two. But I believe we are going to have to do more to provide value in the near future. Let me give an example.

Let’s say the data in the RIS, which is typically in a transactional form and on its own database, is now minable within the EMR data warehouse. Now I have a wealth of data with which I can do some very interesting things, including population health.

I can look at a specific population of patients, and get that data correlated with other EMR data to see how tumor sizes of all African American lung cancer patients in Maryland get affected according to a drug regimen.

We are going to have to ask such questions to bring value out. We’re going to need tools that work together in a more native way so that whatever gets done in one system doesn’t just get stuck there.

Q: Does the app revolution have the potential to help radiology solve its communications and interoperability issues?

Sevinc: Yes and no. There is a lot happening with mobile apps and web apps. We’re starting to have a lot more building blocks that we can replicate and do stuff off of. The challenge is that there really are no set standards on how apps are going to work within the hospital and outside of the hospital. But the potential is certainly there.

Let’s say your expert neuroradiologist is at home sleeping, and there is a special case you need to consult him on. You should be able to send in a quick request so that, with one click on his iPad, he can get to the case and provide his opinion. He should be able to do this securely and without going to a radiology diagnostic workstation which he would have to start up, log in, go through VPN and search for the case to provide the an opinion.

We are going to have to be able to support such workflows and that’s going to likely happen through our mobile devices. Remember that mobile devices are small computers that we carry around with us at all times. We are still ways from using mobile devices for diagnostic-level work, but there’s no reason why we cannot do quick consults and opinions. This is going to help us communicate better and reduce waste.

Q: Are you looking forward to helping to create and fine-tune the solution?

Sevinc: Well, it’s good that the solution is partly here. DICOMweb and HL7 FHIR both have great promise, but are still evolving. Adoption of these new standards is starting. The industry sees the importance of this: Epic is holding a HL-7 FHIR hackathon, SIIM has been holding hackathons for the past two years promoting both. VNAs and PACS are starting to enable DICOMweb.

It is great to see the progress being made and not only watching vendors drive standards, but also having customers ask for compliance for these new technologies and standards.

I am excited to be a part of this revolution. I think it is really going to break the chains on our data.