- Hello, Dr. Kramer, and welcome to our new episode of Limitless Medical Imaging Podcast.
We're happy to have you here.
- Thank you. Happy to be here. Happy to talk about cloud computing today.
- As you've mentioned, I'd like for us to discuss this time the role of cloud computing in healthcare and particularly in medical imaging. My first question is, what would you say about the role that cloud computing is playing in modern health healthcare, and how is it changing the way we deliver and access medical services?
- Well, I think there's a lot of promise in cloud computing, and I think we're hearing a lot of excitement about moving services to the cloud. There are several reasons for it. One is that the scalability and cost may result in lower expenses for health systems that are very concerned about expenses of it.
The second is it allows us new ways to collaborate, and the third is it provides more computing power on demand that we might not have had available to us within our on-prem information systems.
- Could you expand a little bit about the three points, please?
- Sure. As you think about scalability and costs, more and more organizations are growing, adding practices, physicians, health systems, and hospitals; the size of the IT department can't grow in the same fashion. And so we're looking for ways to create an integrated architecture and bring those organizations together, but without creating the increased costs. So mergers, and acquisitions in the cloud are a good use case for considering how to bring those organizations together. I think as you think about new use cases, not just existing infrastructure, but things like AI, we are finding in our organization that we can't run some of the most complicated machine learning algorithms in the fashion that we'd like to run them, in the timeframes that we'd like to run them, with existing on-prem services.
So we're looking for the cloud to provide us on-demand horsepower that we need, both from a storage perspective, as well as a computing perspective for those complex AI algorithms. I think the last here is that cloud computing is helping us think about supporting redundancy and downtime and recovery, certainly around storage.
If we have, say, for instance, an on-prem radiology system, and the cost to create redundancy across multiple server forms is very large. And so we're looking at the storage, both offline and online storage within the cloud.
- Thank you for that. You've mentioned something about a more integrated architecture across our health systems and new models of care emerging with the help of cloud computing, and new digital pathways of care. How is that impacting new ways to collaborate between healthcare systems?
- Absolutely. You know, many organizations are working across, tax IDs or typical corporate structures. We don't really have, and for a number of reasons, it could be because we're creating new, clinical models in a clinically integrated network where multiple hospitals and health systems and practices are under contract. To reduce the cost of care for a population that could be a payer contract, it could be part of the clinically integrated network. It could be a part of various governmental programs. So bringing data together in the cloud across multiple EMRs increases the scalability and ability to collaborate in that clinically integrated network.
A second use case might be where we don't have specific specialties, so contracting with specialties that are outside of our organization or serving across multiple organizations and leveraging a cloud architecture to bring the images and the workflows to them. A good example here is that most pediatric emergency rooms may not have pediatric subspecialists such as a radiologist or a neuroradiologist. And a clinician working with a private radiology company could work across multiple sites.
- Related to the topic of collaboration across health systems. I believe that cloud computing is also improving health equity by allowing patients in rural or underserved areas to receive the same level of care as those in larger cities with more resources through connecting different healthcare systems and through connecting specialists, for example, as you mentioned, pediatric radiologists or radiologists who are highly specialized in a particular, area, discipline, to help areas with more limited resources.
- Absolutely, and we talked a little bit about pediatric neurology, but oncology and tumor boards are another complex use case where specialists might not be located in the location. In fact, you know, in the circumstance of the Ukraine-Russian war Medicai has been able to provide oncology radiology services, across borders and in areas, not just underserved, but under horrific circumstances. So we've demonstrated that use case, particularly where there's a lack of infrastructure, whether it be rural or, unfortunately in an area that's under conflict.
- Yeah, that is true. What would you say about the topic of privacy and security? How does the use of cloud computing impact patient privacy and security, and what steps are healthcare providers taking to ensure that sensitive data is protected while using cloud computing?
- Well, you know, first off, this is not just an issue for the cloud, it's an issue for all of us in healthcare, we
have to be incredibly mindful of how we look at risks to our health system, to our patient's data. So culture and constant vigilance are necessary regardless of the platform that we're on. Malware attacking a system, because of an email that we innocently clicked on, is a horrible thing for a health system and there certainly are plenty of examples of that, regardless of whether or not they're on the cloud. That said, the cloud provides us an opportunity to provide redundancy and backup, under circumstances where privacy or attacks to the health system have occurred. I think, the other aspect of the cloud is that many of our on-prem systems and our on-prem experts are not as sophisticated as we would like.
It is really, really hard to get the chief information security officers and the privacy expertise that we need in every entity. And so we have an opportunity here to leverage a cloud and the expertise of large architectures, such as Azure and Microsoft or Google or Amazon, where they are constantly refining and building an architecture that's very robust.
And so, the cloud architecture gives us an opportunity to leverage that without having to build that expertise locally. Finally, I think there are risks to a centralized model, you know, a single point of failure, whatnot. But, I think I would take my chances on a robust, hardened architecture in the cloud over managing all of the points of failure that we could have within a local architecture.
- There's been a lot of talk about the potential benefits of machine learning and artificial intelligence in healthcare, and particularly medical imaging. How does cloud computing factor into this situation and what sorts of tools and technologies are emerging in this space?
- Right. You know, I think we come back to the initial comment that we made is that, having scalable on-demand storage and compute power provides scalability and capabilities that we don't have on-prem.
You know, getting server racks installed and ramping them up and paying for them on an ongoing basis is not something that is manageable by certainly small entities, but even some of the larger health systems. So the first obvious answer is scalability and computing power. I think when we think about some of these models, even just some of the machine learning models that might have somewhere between 30 and 120 features or variables, getting those to run on an ongoing basis.
And as an example of this is sepsis, our sepsis model, which Epic had helped develop, runs every 30 minutes across a thousand beds and constantly updates. That's not something that we can run on the typical EHR hardware located on-prem. We moved that to the cloud quite a while ago, and it's been very effective for us, but I think that's a simple, maybe not so simple, but that's a machine learning model with a small number of variables.
But if we start to look at image analysis and image recognition, we get into large, generative language models like ChatGPT, and even natural language processing. Our ability to run that against the totality of a four-terabyte EMR such as what we have is just not feasible. So we've gotta move to the cloud and use the scalability and horsepower that can be brought online, rapidly and then shut back down or keep running for ongoing algorithms.
- Thank you. Let's talk a little bit about the benefits of cloud computing in what concerns, the scalability of healthcare organizations. How is this particularly beneficial and what sorts of challenges does it pose for IT departments, in a healthcare organization?
- Sure. You know, we kinda talk about cloud as like, it's set benefit, like you, you immediately get cost savings and, and scalability. But it's somewhat of a holy grail. It depends on the business use-case as to whether or not those have been achieved. So you think about software as a service, very simple, you know, web-based cloud architecture, Salesforce, various other, HR and financial systems, web based, we can get benefits of the cloud relatively quickly.
But you think about other use cases such as managing clinical data, this has been more challenging. So we really have to look hard at the use cases and think very carefully about whether or not a particular use case can move to the cloud.
For example, moving an entire EMR to the cloud, maybe hasn't been demonstrated. A few cases like Athena. But not the totality of a typical, server based architecture. So, there are some examples where the financial cases are emerging, and we talked about AI process automation and managing complex workflows across entities, large scale data integration, and certainly storage and redundancy for large files and large storage needs.
And a great example here is, and, and one of the reasons why Medicai exists is, being able to do radiology and other complex modality file management. And so Medicai is really a file management platform, with storage capabilities and viewer capabilities. And all of those can sit in the cloud and can do so quite affordably.
We've been able to demonstrate that for our clients. This, that we can substantially reduce the storage costs and provide a great deal of flexibility in terms of workflows and and file management.
- As more and more healthcare organizations embrace cloud computing, what sort of skill sets and knowledge will be required for IT professionals to manage these systems and maintain them?
- Absolutely. Traditional computering engineer programs teach coding and teach waterfall type programming skills and have really not yet into cloud-based computing and architectures that are rapidly emerging. It's a skill set that has to be developed, possibly outside of traditional programs.
It's coming. And certainly the educational organizations offer some good programs going forward. But, you know, things like ChatGPT and AI, software as a service versus a platform, all of these things are emerging technologies that your IT department may not have developed. And so, in one case I know of, made cloud a strategic priority and contracted with a third party vendor to put over 200 IT staff through programs of training and education.
So I would recommend the ongoing education and the retooling of your IT workforce, specifically around things like AI, digital, cloud-based computing, and leverage some of the tools available from Amazon, Google, and other third parties that can help achieve certifications and other skill sets.
- Okay, thank you. How does the use of cloud computing impact the speed, and efficiency of medical imaging workflows? What are the opportunities from an architecture perspective, from a workflow perspective, from a user perspective in terms of speed and efficiency?
- Absolutely. From an architecture perspective, the cloud exists outside of your traditional systems.
So once that data and connections are made into the cloud, you now have a whole introduction of open source, AI models, even new data sets. I've seen organizations starting to combine things like weather, sporting events and different situations in a community, such as demographics into their AI models that they, that they would not have normally been able to bring into their on-prem databases. So that's really interesting, to be able to, for instance, to predict ED arrivals, you probably need to know something about what's going on in your community.
- Very interesting.
- But also the ability to bring in algorithms, that we wouldn't normally be able to bring into a traditional. server-based architecture, being able to spin up servers, spin up those algorithms can be done much more rapidly in the cloud-based architecture from a workflow and user perspective.
So something completely different than just technology, but, but being able to deploy cloud across entities very rapidly. Many of our vendors are turning on cloud-based architectures across clients, with relative ease. We're not waiting for servers to be delivered, turned on, configured. And then we're able to bring data, and access information from across the boundaries of a traditional healthcare system.
So, for instance, with radiology imaging, you know, as soon as we know that there is an entity that's coming into a radiology service area, we can give access to a cloud-based architecture, turn on our nodes and start to allow radiologists and other expertise to access that information extremely quickly.
- Thank you so much for these insights, Dr. Kramer. I have a one final question for you. One of the key challenges in medical imaging is managing complex data sets that require a lot of storage and processing power. How does cloud computing help to address these challenges? What kind of infrastructure is required to support these systems?
- This is a good question that kinda summarizes what we've talked about earlier, but cloud can create a common user platform across entities. The combination of applications that run in a web-based environment and cloud host environment with common storage provide tools across entities.
And this is certainly true of imaging. Where the imaging modalities are growing more and more complex, and requiring more and more expertise every day for the purposes of patient care, learning, research, or development of new algorithms, the cloud is uniquely suited to provide greater and greater benefits to those that are implementing those use cases.
Second, that architecture is affordable. Storage in the cloud can be half to 20% of what we are seeing on-prem, and in local organizations. Next, the cloud has access to great number of application programming interfaces, FHIR standards, HL7. Data models and complex data management tools are readily accessible, generalizable, open source, available in the cloud.
Next cloud is going to allow us on demand and ongoing horsepower for complex processing, large data sets, deep learning, neural networks, all of the buzzwords of AI and the reality of AI are coming to life, for use cases that use those tools. We talked a little bit about collaboration for clinical care, but new discovery and research - cloud tools are going to allow us to anonymize data, to de-identify, re-identify, create digital twins, and do research in ways that we never have done before, on a scale that we've never done before. So, using cloud for pharma, say for instance, understanding a population of patients, that have been undergoing chemotherapy, looking at the images and means to detect remission, progression, metastases, whatever it might be. You could only do large scale digital analysis on image, within a cloud course power. So, a lot of opportunities, a lot of use cases. I think we're gonna see the combination of AI large data sets, processing power, all converge on the cloud. And that's gonna be strategic for organizations that want to be leading edge, in terms of augmenting their workforce, reducing costs, and providing greater services more reliably to their patients and to their providers.
- Thank you so much Dr. Kramer, for taking part in this episode and for sharing with us from your experience. Everyone, please send us any thoughts or questions you might have for Dr. Kramer, to email@example.com. Feel free to message us with your feedback on our social channels and follow us on medicai.io. Till next time, thank you.
- Thank you, Andra. Thanks for having me.