When the COVID-19 pandemic hit earlier this year, it presented health systems across the U.S. with a need to accelerate innovation during an unprecedented time that shut down many in-person medical services. Health systems quickly turned their attention to rapidly scaling telehealth, deploying artificial intelligence (AI) and improving revenue cycle management, according to new research from the Center for Connected Medicine (CCM) at the Pittsburgh-based UPMC health system.
The CCM’s fourth annual “Top of Mind for Top Health Systems” report, which surveyed 117 executives representing 112 health care provider organizations, focuses on how innovation priorities shifted in response to COVID-19 and the role of key technologies in managing the pandemic. The report, conducted in partnership with KLAS Research from May to August 2020, was published on October 20.
Survey results indicate these technologies were priorities for the future at many health systems before the pandemic, but COVID-19 caused them to fast-track implementation, looking to telehealth to continue seeing patients, AI to enable better decision making and revenue cycle management technologies to improve efficiencies.
Among key findings, the Top of Mind report shows:
> Nine out of 10 organizations successfully met increased telehealth demand during the pandemic; however, quick implementation of solutions magnified opportunities for improvement, including integration and patient and clinician experience, that many will seek to address in the coming year.
> Three-quarters of respondents said their organizations are measuring and analyzing data from telehealth use; however, a limited number are examining health outcomes related to telehealth.
> Half of respondents reported using AI in response to the pandemic for applications such as clinical decision support, management of beds, staffing and devices, and analytics—experience that is boosting interest in the technology and pointing to greater utilization in the year ahead.
> A majority of respondents said their organizations are using 20% or less of their health care data to inform AI applications, suggesting health systems need to do more to standardize and share their data.
> Identified as an area most in need of innovation, revenue cycle management has become a greater priority for health systems, with 57% of respondents saying they are optimistic or very optimistic that innovation can happen in the coming year.
> Predictive analytics, AI, bots and automation are cited as the most-needed technologies to improve revenue cycle management (cited by 26% of respondents) as health systems look for opportunities to be more efficient.
“Technology has been so essential to the COVID-19 response at UPMC and other health systems that the line now is blurred between traditional health care and digital health. Technology and digital applications that once were not used to their full potential are now a permanent part of providing the best possible care for our patients,” said Rob Bart, M.D., chief medical information officer at UPMC, a founder organization of the CCM, in a statement contained in the press release. “It now is our responsibility to create an excellent patient experience with all of the tools at our disposal, whether they are virtual or in person.”
As the report explained, “The Top of Mind 2021 research project included two surveys. First, an initial survey was sent to health care leaders to solicit responses to the following questions: (1) What area in health care has the greatest need for innovation/disruption? (2) In what area of existing technology have you seen the greatest progress/improvement? (3) What do you see as the most exciting emerging technology in the next two years? Sixty respondents participated in the initial survey. The second, more in-depth evaluation focused on the top responses from the earlier survey and was conducted via phone. Researchers asked both quantitative and qualitative questions to gather further insights into the highest-interest topics and learn how technology trends impact health care organizations’ strategies and priorities. One hundred and seventeen executives from 112 health systems across the United States participated in this second survey. Seventy-nine of these respondents are executives (C-level executives, VPs, etc.).”
With regard to the COVID-19 pandemic, the key question the survey asked was this: “What were your top three innovation priorities prior to COVID-19, and how have they changed?”
Here’s what the survey found:
> Prior to the COVID-19 pandemic, healthcare organizations were focused on business-effiency innovations (such as those related to revenue cycle management) and improving operations.
> Telehealth was also an innovation priority, but many respondents reported only slow movement toward greater utilization.
> When the pandemic hit, healthcare organizations rapidly shifted their focus to enabling staff to work remotely and dramatically scaling telemedicine and virtual care.
> Survey respondents predicted that their top priorities will continue to be affected and hindered by the pandemic, due to budget constraints and uncertainty about the future.
> Respondents also said their organizations need tools such as bots, business intelligence, and artificial intelligence to improve visibility and decision-making abilities—accelerating a need that existed before the pandemic.
So, prior to the COVID-19 pandemic, the top innovation priorities among health system executives surveyed were:
> Revenue cycle management (26 percent)
> Telehealth/virtual care (26 percent)
> Business plans to improve operations/efficiency (21 percent)
> AI/BI/bots (15 percent)
> Population health management (11 percent)
> Integration/interoperability for improved patient care (11 percent)
> Taking on risk/optimizing quality and reimbursement for risk-based contracts (11 percent)
> New technology solutions (e.g., EMR) (9 percent)
> Consumer digital strategy and experience (9 percent)
> Ancillary systems (e.g., cardio, lab, clinical decision support) (5 percent)
What future changes do executives expect to make to telehealth because of COVID-19?
> Continue to use telehealth (17 percent)
> Improve integration (14 percent)
> Expand use of telehealth (14 percent)
> Contract with a telehealth vendor (9 percent)
> Upgrade infrastructure (8 percent)
> Optimize workflows (6 percent)
> Discontinue/cut back telehealth use (5 percent)c
> Consolidate to single platform (5 percent)
> Utilize home medical equipment/remote patient monitoring (5 percent)
> Centralize telehealth departments (2 percent)
> Expand number of physicians doing telehealth (2 percent)
> Unsure (9 percent)
As the report noted, “Respondents frequently reported they are concerned about where regulations and reimbursements will end up after the pandemic slows down. One CMIO shared, “With COVID-19, a lot of regulations have been relaxed, and I don’t know whether that will be temporary or permanent. What has made telehealth much more usable is obviously reimbursement changes. The other change is reciprocity between states for licensing; for example, if a provider holds a license in one state, regulations state they can’t necessarily conduct a telehealth visit with a patient sitting at home in another state. Those regulations have been relaxed because of COVID-19, in general. Whether that will continue remains to be seen.”
COVID-19, artificial intelligence, and AI for COVID-19-based planning
The second major area of the report concerned artificial intelligence and its application to the COVID-19 pandemic. Respondents were asked whether their organizations were using AI to help manage COVID-19; how COVID-19 has changed their approach to AI; in which areas AI is being applied; what percentage of their organizations’ data is being used develop and inform AI solutions and applications; and their perceptions of the current level of AI regulation in healthcare.
The key findings were thus: “Half of respondents report using AI technology to help respond to and manage the effects of the COVID-19 pandemic. Interest in AI has increased since the COVID-19 pandemic began. Health care organizations are most often leveraging AI technology for clinical decision support (CDS). Revenue cycle management is the top area where organizations are planning to leverage AI in the future. AI technology is most often being implemented in the form of vendor software solutions; it is less common for healthcare organizations to build their own AI capabilities. More than half of organizations use less than 20 percent of their health system’s data to inform AI solutions and applications. A majority of respondents are satisfied with current AI regulations; however, many are concerned that future restrictions around privacy and security will make it more difficult to use AI solutions.”
Some of the details are particularly interesting as well. Asked whether they were currently leveraging AI for various types of activities, or planned to do so, the following were results, with the level of current use first, followed by planned use:
Ø Clinical decision support: 61 percent/19.5 percent
Ø Dictation assistant or transcription: 50 percent/21 percent
Ø Diagnostics for medical imaging: 48 percent/16 percent
Ø Quality reporting/improvement: 45 percent/14 percent
Ø Financial modeling: 43 percent/14 percent
Ø Revenue cycle management: 35 percent/38 percent
Ø Cybersecurity: 35 percent/14 percent
Ø Real-time triage: 35 percent/24 percent
Ø Personal health coaching: 25 percent/5 percent
Ø Fraud detection: 24 percent/28 percent
Ø Drug discovery: 22 percent/6 percent
Ø Virtual assistant: 18 percent/41 percent
Ø Genomic analysis: 11 percent/22 percent
Experts weigh in
Following the publication of the report, Healthcare Innovation spoke with Rob Bart, M.D., CMIO at UPMC, and Jennifer Despain, KLAS’s director of market analysis, to obtain their perspectives on the results of the survey and the report’s analysis. Below is a composite of excerpts from those two interviews.
What was your overall impression of the telehealth-related findings in the survey? Fifty percent of those surveyed are currently using AI for COVID-19-based planning purposes.
Jennifer Despain: We’ve seen the benefit of telehealth, in terms of reaching patients that might be harder to reach, and the many use cases for telehealth. So I’m optimistic, and I think providers are optimistic, that the government will maintain some of the relaxation of regulations for telehealth, going forward. And hopefully, the reimbursement will stay at a level where it’s profitable for patient care organizations to do telehealth-based care delivery.
Rob Bart, M.D.: The focus on telehealth, given what’s happened over the last six months, that part didn’t surprise me. But I did think that some of the findings around supply chain and around working from home, were interesting. In healthcare, we don’t traditionally think about working from home; but with the movement towards virtual care, there have bene opportunities. And certainly, an organization like UPMC has invested a fair amount in human infrastructure, so supporting those individuals has been very important to us, and I think that other organizations have also recognized that.
What about the findings around providers’ concerns over whether the Centers for Medicare and Medicaid Services (CMS) will continue parity-based reimbursement for telehealth-based care delivery?
Despain: Clearly, that was the biggest concern. But I think the government will keep it at a level that will work for providers. And one area of concern will be around security-based regulations. We can’t have providers Facebooking their patients.
Looking at the use cases for artificial intelligence, what strikes you about current versus near-future leveraging of AI for various use cases?
Despain: I think that one of the reasons that revenue cycle management has such a big spike, as well as virtual assistant, for future applications, is because of COVID. Organizations will need to be able to make better decisions based on data. That’s one reason people want to incorporate AI into those areas. And I think, per genomic analysis, that’s still a pretty new area. That’s not been widely adopted yet. And then CDS, so a lot of that is NLP-type AI. They’re able to sift through populations to find at-risk patients they can help—so NLP has already been widely adopted. A lot of these CDS tools have already incorporated AI into them, so they’re automatically using them already.
Bart: I just interpret it as the early growth phase in the adoption of AI in healthcare. I don’t necessarily interpret those who haven’t yet leveraged AI as laggards. Let’s look at bed management, for example: there are traditional algorithms you can use to manage beds, but there are also machine learning-based algorithms that can be used. So, is the use of AI or machine learning a requirement for a particular circumstance? And right now, we’re still in the early phases of the adoption of AI, and there are early adopters who say, I absolutely want to use AI in the management of my bed capacity; and there are other organizations that continue to use traditional algorithms well. So I see it as sort of an arc of innovation, and there’s a large range of where it’s healthy to be on that curve of adoption. Certainly, there aren’t too many organizations either at zero percent or 100 percent, so everybody is moving along that curve of adoption.
What were some of the most interesting findings for you?
Bart: The focus on supply chain—I don’t think of supply chain as being something top-of-mind in late 2020; but certainly having worked through the early and ongoing phases of the pandemic, mastering supply chain has certain been important to the success of UPMC, and it deserves attention in making sure there’s efficiency and effectiveness there. But at a regional and national level, I’m not sure it’s been managed as well as it could be. There have been reports that there was competition for PPE between and among different states, which created more expense or cost in the system. So this is sort of a wake-up or lesson for our country, such that if there’s an event like this in the future, it can be handled in a more efficient and equitable manner.
What does the finding on data analytics for telehealth say, about the leveraging of data analytics, health system-wide?
Bart: When I see that finding, I think that utilizing the telehealth data for clinical operations, analyzing it in detail, that data is already readily available in most organizations, certainly at UPMC. But when it comes to health outcomes data, that is more closely related to chronic disease management, related to the specific diseases that individuals have. Absolutely, the outcomes from telehealth, need to be studied and compared in terms of outcomes from face-to-face care. Take for example the adults managed on a certain cadence via face-to-face care, and compare say, twice a year face to face, versus once a year face-to-face, and twice a year via telehealth, what are the outcomes? And it takes some time and thought to develop such data. And so we’ll need to find out whether there is or is not a disparity in terms of telehealth-based outcomes, compared to face-to-face outcomes. And there will be different cadences in areas such as congestive heart failure, or care for healthy moms. So we’ll have to become more sophisticated in using telehealth and applying those strategies to care and care management.
How do you see the next few years in terms of the ongoing evolution of AI in patient care organizations?
Despain: Providers are going to be finding ways to become more efficient and make better decisions. They’ll try things, but if something doesn’t make them more efficient, they won’t keep them. If AI for imaging diagnosis doesn’t make physicians more efficient, it won’t last. I think we’re optimistic about what outcomes we can get. And if they can do well by them, they’ll go back to less expensive methods. And I don’t think AI will necessarily replace people, it will just support them.
Bart: At UPMC, we’ll be integrating more AI and machine learning into care delivery. At first, it will be more on the clinical operations side; in terms of managing the clinical cadence of disease, that’s still something that’s a bit few and far between. I think that healthcare as an industry will need to become more comfortable with that phenomenon. But in the next 24 months we’ll start seeing more opportunities and evidence in that area.