In order to address transformative issues in the healthcare industry, healthcare provider organizations, payers and life sciences companies need to be agile, need to drive disruption utilizing technology, and need to be more analytically driven, according to a panel of healthcare industry leaders. During a panel discussion at New York City-based consulting firm KPMG’s Annual NY Health and Life Sciences Summit, which took place at KPMG’s Manhattan office, healthcare industry thought leaders tackled the impact of disruptive technologies, such as intelligent automation and robotics, in healthcare.
Joe Parente, principal at KPMG, kicked off the summit by offering a general overview of the current moment in healthcare and the role that intelligent automation, and even artificial intelligence (AI) and machine learning, can play to address the most pressing challenges in healthcare. Parente noted that healthcare provider organizations are challenged with tremendous operating pressures with rising costs and shrinking reimbursements. “What will tomorrow’s successful healthcare organization look like? Agile,” he said, noting that adaptability will be critical moving forward.
“This is a time of uncertainty, with regard to legislation, regulations, increasing competition, driving the focus on cost reduction, along with an ongoing shift from volume to value, and improving quality of care and optimizing reimbursement,” he said. To face these challenges, healthcare organizations will need to see disruption as an opportunity, not a threat, and should drive the disruption, he said. “Organizations can use analytics to find variance in clinical process, re-engineer those processes and then use automation to be more efficient,” he said.
During the panel discussion, Lita Sands, an independent and industry consultant, Dusty Majumdar, chief marketing officer at IBM Watson Health and Paresh Shah, M.D., vice chair of quality and innovation in surgery at New York City-based NYU Langone Health gave their perspectives on the current adoption of intelligent automation, robotics and AI in the healthcare space.
Providing the vendor perspective during the panel discussion, Majumdar with IBM Watson noted that AI in healthcare is “red hot.” “In terms of where we see traction, we see a significant amount of traction in imaging. Radiologists, in our experience, they want something like Watson to help them detect that legion or to characterize a nodule that’s in the early stage of cancer. We do see it across imaging, see it across oncology, in terms of being able to stratify the patient with the right clinical trial, and we see it in genomics in terms of linking the mutation profile to the right clinical trial and the right treatment and also some advanced ways of looking at the genomic profile,” he said.
Majumdar’s comments come a week after Stat News published an investigative piece on IBM Watson’s shortcomings, contending that the artificial intelligence supercomputer as not lived up to its potential. As Healthcare Informatics Managing Editor Rajiv Leventhal noted in a news story about the Stat News report, the piece examined Watson for Oncology’s use, marketing, and performance in hospitals across the world, from South Korea to Slovakia to South Florida. The in-depth article included interviews with doctors that have deployed Watson at prominent healthcare institutions as well as other healthcare experts—many of whom seem far from thrilled about the results Watson has delivered so far.
Offering the healthcare provider perspective on where intelligent automation and AI are gaining traction, Shah said, “At the very granular, foot-solider level, that’s where it’s the hardest to get some traction because they are so consumed with the day-to-day of what they are doing. Interestingly enough, when a provider sees a tangible benefit, they are ready to adopt it. The hurdle is just getting it to them, to show them that it’s added value. Once they see it, they will take it on,” he said, adding, “Things like decision support, such as, if I’m a radiologist sitting in front of a screen, and I see something, if I could push a button and have it verify what I’m seeing, or push a button and have it give me three ideas that I haven’t even thought of, they love that.”
Shah continued, “Or, if I’m an internal medicine doctor, and I may or may not know what is the best option for treating diabetes for this particular patient. When I graduated medical school, there were only four drugs, outside of insulin, that were used to manage diabetes. Now, there’s a couple of dozen drugs. One of the challenges is that, at the provider level, the actual quantity of the information has expanded so exponentially giving providers a tool or vehicle in which they can access, categorize and curate that information, that is really valuable.”
The panelists also acknowledged that organizations need a structure around technology adoption in order to effectively prioritize projects. During an earlier keynote address, Shah discussed NYU Langone Health’s plans to leverage intelligent automation and robotics, particularly in its new Kimmel Pavilion, which will feature TUG robots to distribute food and medications, digital wayfinding tools for patients and families and Brainlab, an audiovisual medical communications system for the surgical and procedure rooms. He also detailed some of the health system’s deep learning IT initiatives to utilize data analytics and machine learning to reduce clinical variability and improve quality of care while reducing costs.
“For us, it started with our transformation to value, that was the primary driver and the mission that enabled us to prioritize different projects,” Shah said. “It was really around moving from volume-based to a value-based system. And, the first simple pass of that was around cost reduction, waste elimination, with the objective being wanting to improve care quality and reduce the cost at the same time. From an operational perspective, that’s how we looked at it. Some of that was easy and some of that turned out to be quite hard, and required some really in-depth analytics to understand it. And now we’re moving to the next generation of quality improvement, which is thinking beyond just what we currently do better, but thinking about how we can actually change how we do it.”
Shah provided an example of how the health system’s Value-Based Management (VBM) program, which serves as the oversight for many of these projects, leverages technology to address problems. Health system leaders found that many hospital patients are readmitted because they get dehydrated. “The current mechanism to handle that in our system is those patients come through the ER, because in the physician office, due to regulatory issues, you can’t start an IV, even though the physician’s office is in building with the hospital,” Shah said. “So, in order to understand how big the problem was, we used our data warehouse and we applied intelligent analytics to look at what the scope of the problem was, not just episodes of reoccurrence, but looking at what the true cost was, what that translates into, and how do we act on it?” The insights derived from that analytics work motived the health system to set up a process to provide patients with hydration in the outpatient setting to avoid patients being readmitted due to dehydration.
As Healthcare Informatics’ Leventhal addressed in a recent article on the opportunities for artificial intelligence in healthcare, one of the key concerns among clinicians is that AI and machine learning will eventually replace doctors. IBM Watson Health’s Majumdar addressed this issue, referring to it as a “myth” around AI. “Our belief is that it will augment humans. If we can augment the oncologist, with the right data at the right time, based on 7,000 cancer-related publications per day, based on the explosion of genomic data, based on all the different publications around treatment that are out there and around drugs, we think we can augment the oncologist, the pathologist, the surgeon to take the right steps.”
He added, “There are some more subtle challenges in radiology. It’s not just about recognizing a tumor, but the characteristics of it. We’re just beginning to get there with machines; it’s going to be a long time before a machine can actually replace the critical judgment of the radiologist who has seen 10,000 similar tumors. You need a lot of data, a few million tumors, for the machine to learn from. We believe that we are, at this point, augmenting radiologists, oncologists, in a big way with cognitive intelligence, what we call cognitive capabilities, and this is a journey. In my opinion, it’s at least a couple of decades’ long journey before we can think of machines completely replacing physicians.”
Shah also noted that adopting technology with cognitive capabilities is as much as process issue as a technology issue. “It’s a process, and an evolution. As you build trust in this technology and as the technology demonstrates enough reliability to engender that trust, you will allow it more freedom. We’re not at the point of cognitive independence; and nor are we trying to get there. We’re trying to maximize the existing potential. With radiologists, for example, if you get yourself lulled into this sense of confidence, or even complacency, around what it’s spitting out at you, you run the risk of propagating error, rather than reducing error. That’s true to everything we’ve done to now. That isn’t new to AI. That’s a process issue more than a technology issue. I think you build processes to protect against that, and you train and educate around that.”
On the subject of disruptive technologies in healthcare, the panelists also discussed the impact of new entrants into the market, such as Verily, Amazon and Apple. Sands, a consultant in the life sciences space, says there is an opportunity for these technology companies, as well as payers, to leverage technology to change consumers’ behavior and, subsequently, have a profound impact on population health. “You have some payers trying to reinvent itself themselves. For example, the CEO of Aetna has gone out there and said, ‘We’re going to take the perspective that we need to move from the exam table to the kitchen table.’ You probably read they made a big investment with Apple watch, using this data to help people change behavior. That’s a great example of a player that’s being innovative.”
She continued, “And then of course, we’ve been reading a lot about Amazon, and I think they call their division 1492, and they are going into the pharma space. With Amazon moving into the space, I’m enthusiastic about it; it’s an opportunity to remove barriers. You could extend it into all sorts of avenues in terms of being able to do clinical studies in the home.” What’s more, Sands said she sees opportunities for technology companies in the life sciences and pharma space to put together claims data with behavioral data, “which is where the nirvana is. “Once you have that, and then use AI, you can now start to create tailored, clinical pathways and we can quickly monitor and adjust a medication. It’s been proven that with access to data, insights and action-driven behavior, you can dramatically change your health,” she said.
Moving forward, Sands says life sciences companies need to change their strategy and become more analytically-driven to succeed in this new, disruptive world. “We’re moving into the world where data is the new currency; data and insights, we have to come to the table with that, it’s not just our drugs, it’s about the whole patient. If that’s what the model looks like and we’re discovering that personalized data from technology impacts human behavior, let’s go to the beginning; let’s start with zero and change how we do our clinical trials, augmented with technology, let’s not be afraid of this,” she said.
The panelists also offered their perspectives on lessons learned from early adopters of these technologies.
IBM Watson’s Majumdar said partnership is a key to success. “IBM doesn’t do anything on their own in healthcare,” he said. And, he also noted, “As we are applying AI to imaging, we have to remember that the imaging technology is not sitting still, it’s also evolving. The MRI machines, and each of these advancements in CT or MRI, are improving the specificity, the accuracy of diagnosis. The key is to tie advancements in AI with the changes and advancements in imaging.”
Shah reiterated the need for provider organizations to have a strategic plan to help prioritize technology-driven projects. “The second lesson is, fail fast,” he said. “One of the biggest challenges is that there is so much out there. Provider organizations are inundated with vendors, of all sizes and shapes, giving you these sexy, appealing ideas, but you’re going to engage with something, make sure that your pilot is able to demonstrate its true value quickly, and if not, then cut and run and move on.” Shah also added, “You have to a have a clinical champion. It can’t be something that’s dictated from on high from the C suite. There needs to be a ground level clinician who wants this to happen; because that’s the person who is going to help you with change management and help you with the clinical case as well as the business case.”