Limitless Medical Imaging podcast, ep. 2: The Importance of Radiology Informatics, with Dr. Mike Kramer


- Hello everyone. And welcome to this new episode of our podcast. Today I will be talking to Dr. Mike Kramer, who is a clinically trained health system leader with a background in IT, quality, and clinical change. Dr. Kramer has led informatics teams for OhioHealth, SpectrumHealth, and TrinityHealth, three major healthcare organizations in the U. S., and several other large integrated delivery systems within the US. He is a Chief Executive Officer at Health Value Leadership, where he is providing clinical informatics and health IT strategy, workflow, and change management consulting to major health systems and IT organizations. In this first part of our conversations with Dr. Kramer, we're going to talk about the importance of radiology informatics.

Hello, Dr. Kramer, and welcome to Limitless Medical Imaging Podcast. We're happy to have you here. - - Thanks, Andra. Thanks for having me on your podcast. Happy to be here.

- I will start by asking a very broad question, question, but, we'd love to know, what are some of the big issues in healthcare that are interesting for you right now?

- Well, it's been a busy three years, and there's been a lot of change in technology, using technology to address COVID. But we've now come out of the worst of it, and we've got some pretty big issues to address in health systems across the U. S. First and foremost, the economics of healthcare are changing quite a bit. Inflation, workforce challenges, clinician burnout, all of those things are top of mind as I network. And, spending time with executives across the U. S., I think we've also got some new technologies and emerging consumer threats. You know, yesterday I signed up for my prescriptions through Amazon and that's a significant change- those sorts of consumer-driven health systems are going to change the marketplace.

So, you know, that puts a lot of pressure on health system leaders, and it makes us think about all of what we're doing. We want to start using automation. We want to use cloud. We want to use AI. All of these things are, again, a response to threats in the healthcare ecosystem, but also those threats are a pressure to do things in a way that maybe is more efficient.

And that's kind of exciting. It's because the window is open to bring these technologies and finally implement them into the healthcare ecosystem,  rather than just throwing more people at it.

- Okay. Could you tell us a little bit more about customization in healthcare?

- So traditional players like large health systems really provided a model where you had to schedule an appointment and you might have to have been referred to see a physician. Consumerism provides access to healthcare services outside of the context of large healthcare systems. And so if you're a health care system leader, that's pretty challenging. It levels the playing field. It's no longer an insurer- health care provider- payer type of relationship and transaction. I mentioned briefly that I went on to Amazon and signed up to get my prescriptions through them. They're using economies of scale. They're large distribution networks to take a prescription that previously cost me $20 and reduce it to 5. So, consumerism has a huge impact on the cost and value of care.

Amazon, Walgreens, Village Medical, One Medical, those are all examples of non-traditional consumer-driven health care organizations. So that's one area that health systems are concerned about. The second area is access to information. Consumers can own their own information and with the 21st Century Cures Act, electronic health records information had to be available in 2021.

A limited data set at that time, and then as of October 6th, 2022, all of the electronic health information now is accessible to patients either through a variety of means through FHIR interfaces, through patient portals, basically giving access to information in ways that that consumers never had before.

So, what do you do with that? Well, you own it. You have the ability to bring it to other caregivers. You have the ability to move from one health system to another health system. You can use it for advanced purposes like population health, disease management, research. So those are pretty exciting consumer trends that we're seeing.

And I think one of the things today we wanted to talk about was radiology informatics. And so what happens when you free up images that have information in them that maybe I could use or share with others that maybe weren't initially involved in my care? So that opportunity now exists because of the 21st Century Cures Act.

You know, as you think about interoperability, who needs to have access to your information? Historically, that's been extremely challenging. You know, we're not the bank where you could go to any ATM.  Information and transactions about health care are very complex. There's a lot of varied vocabulary, but really, in the last 10 years of meaningful use, that's been solved for textual data.

And it's very easy to move a document now using HL7 standards, where you can say, this is a discharge summary, this is an admission note, these are my problems, these are my meds. All of that is textual data and the challenge now is imaging data and how do you have interoperability of imaging data.

And of course, the imaging data is getting ever more complex, larger file sizes, different disciplines, neuroradiology, cardiology, cardiac radiology, many more files and types of modalities now growing ever in size in the radiology space. So again, interoperability has been great for text. Now we've got to address some of the more complex files and imaging types, for interoperability.

- And how is this related to radiology informatics or how does the CMIO make sure that they can create a radiology informatics strategy that addresses these challenges?

- Absolutely. You know, I think traditionally back in the day, radiology and the technology around radiology, the goal was to convert the old films that you would hang up on a lightbox to a PACS imaging system.

And if you're really good at that, you saved it in a central radiology system across health systems. And you could go across hospitals, and you could read from anywhere in the world. That was great. And then you started getting into archiving, into what we call vendor-neutral archives. And that was the science. So, the science of radiology informatics was about, you know, the acquisition, the storage, the transmission and interpretation of radiology images.

And what we're finding now is that there's a whole other layer of complexity on top of this, which is that we're starting to do artificial intelligence, we're looking at the images in an automated fashion, we're creating decision support alerts that depend on AI algorithms, and some of those algorithms aren't all that clear to what's happening.

These are deep neural nets and learning algorithms, and the complexity behind these things is significant. So what I've found is that I really need to elevate the role of radiology informatics in our health systems and have them help us understand what's possible. I'll give you an example.

There are some major manufacturers who are detecting strokes earlier in the process of acquiring the film, immediately after the film (sorry, I say film, but it's the file is acquired from the CT scan), within minutes and then notify the stroke team. Algorithms are being brought to market every day in every category of potential diagnoses and are going to be, used in all of the major health systems.
We need radiology informatics to help us understand that complexity, maintain it, make sure that we're purchasing the right solutions and moving forward with it.

- Thank you for this example. Could you give us other examples of use cases for radiology informatics? - Yeah. So, you know, advanced radiology and radiology informatics, you know, what are the problems we're trying to solve? The imaging files are getting bigger and bigger. So you might have a gigabyte to do a digital mammogram. That wasn't true two to three years ago. So the size of files is growing extensively. How do you manage those files, figure out what needs to be in hot storage, which is readily accessible, available for comparisons, versus deep storage, which might be on a static medium?

So that's a use case - archiving and cost-effective archiving. So now once you've got all these images, and let's say you've moved them to a cloud-based architecture. Am I organizing them and storing them and creating metadata that allows me to retrieve those in an intelligent fashion? That's radiology informatics- is the semantics and the metadata around DICOM and non-DICOM images.

If you think of it now, I've got a really well-organized cloud-based storage. Can I do something more with those? Can I integrate across larger health systems? Can I bring in collaborative networks where maybe a specialist in a certain type of disease doesn't exist in my organization? Pediatric radiology is classic. You know, a lot of emergency rooms see peds. But they don't have pediatric radiologists on call. And so, is there a way to bring those radiologists who might be in another organization? And we've done that for a long time, but a cloud-based architecture allows that to be more dynamic and tap into a broader network.

So collaboration and image sharing become a major use case, particularly for subspecialty radiology imaging. As we think more and more about AI, how do I get images out into an area where I can do the learning and analysis of those images and create new knowledge and discovery? So that's a use case.

And the challenge behind that is how do I anonymize those images and use those in a way that I can then create algorithms that can then be brought back down to the health system? Some of those things aren't feasible without radiology informatics, without advanced analytics like cloud-based image management, metadata management, and archiving.

- And without interoperability standards.

- And without interoperability. If I am a researcher or I'm in marketing of new pharmaceutical, can I use these images now that are in the cloud, another use case, right, for the development and long-term outcomes assessment of pharmaceutical development in the industry?

And again, you know, I've got to be able to safely manage the data, the confidentiality. But then re-identify those images, and possibly use them at the bedside, to show what the possible outcomes are for those patients.

- Wow, this is fascinating. How do you think all these, how do you think radiology informatics as a discipline contributes to the overall efficiency of radiology departments?

- Sure. You know, radiology departments are getting overwhelmed with a variety of files, file sizes, modalities, and radiology informatics and new technologies in radiology informatics, are needed to help us manage that complexity. So you think about a tumor board, somebody is going to think about precision medicine. You might be looking at genomics in the EHR, but at the same time, you're looking at a CT scan, and you might also be looking at a nuclear medicine study, and then understanding what the patient's cardiac function is. All those modalities coming into one solution, so that a multidisciplinary team can develop the treatment plan.

That's very complex stuff, and our radiology informatics and our technologists would be thinking about what are those workflows for a tumor board. How do I bring images from across multiple different platforms, multiple different health systems, so that we can draw conclusions from those efficiently?

I can think about participating at tumor board when I was training, and how we were scrapping the other suitcases of information. The patient was bringing information to us. We had CD ROMs. We were lucky if we had a complete view of the picture, and I think increasingly tools that can be agnostic to a health system, to a PACS system, to a discipline, whether it's radiology or cardiology, as well as integrate with the electronic medical record, are needed to get a complete view of the patient.

Andra, you asked me about, you know, what radiology informatics is and what might their function be in the future. And you think about, you know, the old PACS system and making sure it was quick and running.

But increasingly, we're going to have to address these other use cases: What is the lowest-cost archiving solution? How do I bring in images across networks of providers? What do I do in a joint venture? How do I manage referral information? I'm part of now a clinically integrated network or I'm a private equity firm where I'm supporting multiple hospitals, new ones every day. And so a radiology informatician is going to have to be thinking about all those business cases.

And then if you're part of an academic organization or a research organization, you have to be thinking about tumor boards, research, and anonymization of that data. And, then if I'm working in the space of just trying to manage consumer requests, how do I release information to the patient?
That is a huge scope of work, and as a CMIO, I can't do that by myself. I have to ask experts to bring together technologies that have never been brought together before.

- How does the implementation of radiology informatics impact the role of radiologists and other healthcare professionals?

- So, the discipline of radiologists within an organization as a specialty has been about moving through lots of studies and responding to the questions that were ordered of them, and then being a consultant with the treating physicians. A lot of that is brute force trying to get through as many possible images as they can. At what point do they think about their workflow? At what point do they think about new technology? At what point do they think about some of these use cases that we've talked about? And they're going to need experts such as radiology informaticists to help them apply technology.

So, I had a private equity firm come to me about a year and a half back and said: "we're looking at AI solutions. Can you help us implement that?" And there were multiple multiple AI solutions. None of them ran on the same platform. All of them required separate feeds out to the AI cloud, and it's going to require radiology informatics professionals to manage that on behalf of the radiology profession.

- In what ways do you think radiology informatics supports the transition to value-based care?  - Absolutely, you know, I think, increasingly, imaging modalities were associated with one health system and were to address one problem and in value-based care we're going to have to be efficient and effective at using expensive radiology modalities. Organizations like the centers for Medicaid services are looking at the total cost of care. So, my relationship with a radiologist is going to be: what is the first best study? Are you able to access studies that have already been done? Can you layer upon that more intelligence that might help me manage the life of the patient?

An interesting example, I was asked to comment on a company that has an AI algorithm that is able to, from a plain film, so the question might be: does this patient have pneumonia? But the AI algorithm runs and can detect osteoporosis from a chest x-ray, almost as good as a bone scan. Well, you know, does the radiologist embrace that and say, instead of asking for another study, can I use that AI algorithm to more cheaply and more effectively direct that patient to cost effective care and treatment?

- Because we're heading to the final minute of our episode, I would like to ask you, what do you think is the future of radiology informatics, or where do you think it is headed? What trends do you foresee coming?

- Thanks, Andra. You know, I think interoperability is going to drive a lot of it.
The opportunity to have a consolidated transportable imaging record similar to your textual records is coming, to be able to move that to the cloud and be able to direct that both from the treating and managing clinicians to the consumer. And you think about, you know, a young child born with a congenital health defect. Tetralogy of Fallot, for example, which requires multiple, multiple surgeries early in life and the anatomy of that study is going to be captured in many different types of modalities. It's going to be x-rays, it's going to be CT scans, MRIs, you name it from across the hospital that delivered the infant to the tertiary care center that is going to manage those surgeries from the early life of the patient through to adulthood.

Well, that repository information being owned by the patient and being available throughout the life of the patient is where we're going. And I think the opportunity then to apply A.I., to normalize the data and the metadata to keep it efficiently in that cloud architecture is coming and I'm excited to see the day where I, as a parent, or I, as a primary care physician, can confidently say that nobody's going to miss important information and that we're going to have it managed in a way that we're not repeating studies because we can't get access to them, and that we're going to learn things about patients that we wouldn't have otherwise learned.

So all of that's coming, and it's going to depend on large compute, large cloud-based architectures with large compute capabilities and with creativity and the wisdom of the radiologists, clinicians, and radiology informaticists managing that architecture.

- Well, thank you so much, Dr. Kramer. Thank you for taking part in this episode and for sharing with us your experience and some great insights. We'll be talking more about AI and the implementation of AI and its connection with radiology informatics in a future episode, but thank you for your time, Dr. Kramer and yeah, looking forward to our next episode.

- Andra, thank you for having me. It's been a pleasure!

About the author - Andra Bria

Andra Bria is a marketing manager at Medicai. She is interested in health equity, patient experience and value-driven care pathways. She believes in interoperability and collaboration for a more connected healthcare industry.