How AI Use Impacts the Entire Care Team

^{By Julie Gould}

Emad Rizk, MD, president and CEO of Cotiviti, highlights implications of artificial intelligence (AI) that he recently discussed with Congress and the White House Office of Science and Technology Policy and explains ways AI will impact payers, patients, and providers.

Please tell us a little about yourself.

My name is Emad Rizk, and I'm the President and CEO of Cotiviti.

I've been in healthcare for over 30 years on both the payer and the provider side. One of my passions is to continue to bring payers and providers together, create interoperability, and leverage technology in healthcare.

You recently joined members of Congress and the White House Office of Science and Technology Policy to discuss implications of artificial intelligence at an event hosted by the Bipartisan Policy Center, what are some interesting takeaways from that discussion?

First, I just want to compliment the Bipartisan Policy Center for continuously fostering this level of collaboration among providers and payers in both the public and the private sector. They do a very good job of continuing to bring all these stakeholders together.

Artificial intelligence has been around for decades and has been leveraged by multiple industries. In the past five to ten years, there's been an increased adoption of artificial intelligence in healthcare.

Some of the key takeaways are, number one, that leveraging data is a priority for both the government and the private sector. But even though we have experience with artificial intelligence in finance and various industries, we really do need to begin to leverage artificial intelligence and interoperate clinical and financial data in healthcare, because we believe that it will truly benefit the industry in a big way.

Number two, this is a very good time to leverage AI in healthcare because we're now more electronic and digitized than we were a decade ago. In any type of technology, you want to make sure that you digitize as much as you can, and so I do believe that healthcare is in a good, pivotal moment in terms of leveraging the technology.

The last and most important takeaway is that we must be cautious as we leverage all these technologies. We must make sure that the data is complete, that we're continuing to enforce security and privacy, that all the algorithms that we leverage are appropriate and treat everybody equally, and that there's no level of bias that exists across any of the technology that we deploy in healthcare.

How will your work with the Bipartisan Policy Center foster collaboration among providers, payers, policy-makers, and other stakeholders? How will this collaboration be uniquely beneficial in implementing AI in the healthcare space?

The Bipartisan Policy Center is not just a think tank that brings all these stakeholders together to think through solutions— they work through a level of detail for deployment. They bring private and public thought leaders that are experts in their field across multiple industries. This level of cooperation, across industries, segments, public and private leaders fosters significant cross-pollination of information and key learnings.

That's the best way to move forward because there are limitations to the private sector and what we can do, and there are limitations to the public sector and what they can do. Any outcome between the two, must really be through education, understanding on what the obstacles are, and implementing the appropriate technology solutions.

The Bipartisan Policy Center is going to bring key stakeholders together. We are going to work and educate each other on the issues and the data we're going to use, and ultimately the best way to deploy solutions that will improve the healthcare system.

What are some ways AI will impact payers, patients, providers, etc?

If we were to look at AI in healthcare, what we want is to be able to do two major things. Number one is to improve the quality of healthcare across the entire population. Number two is reduce the level of inappropriate care, waste, and cost.

When we look at AI, there are these two major levers. We know as a country we could improve upon the quality. We know that we miss certain quality gaps with some key diseases that afflict many of our patients and our citizens.

How do we leverage AI to basically close those care gaps? How do we make sure that the diabetics are getting all the care that they need, that they're getting all the diagnostic work that they are entitled to? And then, how do we reduce inappropriate care, mistakes, and reduce administrative burden, waste, and fraud?

We are really looking at a point right now where we spend a significant amount of money in healthcare. We have a great healthcare system. We've now digitized a good percentage of it. We still have to get more, but we do still have some unstructured data. AI is helping us take action on information that was previously dormant.

As we get that data structured, and we bring financial and clinical data together, and we begin to drive interoperability across all of those data sets, I believe that AI will take us to the next level of quality, and to the next level of cost reduction.

What are some of the risks of using AI-based tools?

There's been a few articles that have been published around the risks of using these tools. Like the adage, basically you get the outcomes based on the quality of the data that you leverage.

If you leverage a certain level of data – let's say you leverage just cost, or just demographics, or just clinical and utilization information, each data set will yield a certain lens of outcomes.

For example, if somebody has not used the healthcare system for the past two years, you could potentially extrapolate that the individual is healthy. That may not entirely be true because that person may not have sought care when needed or that care was provided outside of their health system.

You could see that the level of completeness of the data and the quality of the data is very important to be able to have good confidence in the output of that data. The biggest risk is that you get incomplete data, or the quality of the data is not accurate.

How do you ensure the quality of the data you’re using in AI tools?

You must make sure that there is a strong quality assurance and focus on three things that are important.

First, ensure the data that you're getting has been quality-proofed and that it is accurate data. I think that is very important that you make sure that there's a level of audit that occurs around the information that you're getting.

Second, quality is not just around accuracy, but it's also around the completeness of the data. What you do not want to have is partial data. You can potentially have a partial result of an x-ray, without the result of a CT scan, or a blood test. Without all the data, you will not get a full picture. The technology platform or lake needs to be sophisticated enough to leverage a complete picture of the data that we're collecting.

Third, you need data that comes from multiple sources. You need data that comes from claims because that will give you the impact of leveraging the healthcare system through the retrospective claims history. The technology needs to have 360-degree view of a patient, financial, clinical, and demographic information.

There are many studies that show demographic information is just as important as genetic information. It can be an important addition to claims and clinical information. 

Is there anything else you would like to add?

I think that the healthcare industry has sometimes been a late-bloomer in adopting technology. We've lagged behind other industries. We now know that technology has helped many industries become very efficient and secure and has reduced a significant amount of administrative burden and has made our lives better.

Whether it's in telecommunication, finance, or manufacturing, we've seen technology make our lives better.

I do think that healthcare right now is at the best point where we can leverage electronic medical records and all digitized data.

As we have become much more digitized, year over year, this ensures seamless and more reliable data inputs, which creates a better output.

The last thing is that we must ensure, as we adopt all these technologies like AI, machine learning, and natural language processing, that we continue to be very disciplined around security and privacy.

We should always make sure that whatever output and whatever technology that we leverage in healthcare is secure and preserves the privacy of the individual and of the patient. We have to always remember that when we deploy and utilize AI we are the stewards of the data, but more importantly, of people’s trust in the healthcare system.