AdventHealth’s Genomics Risk Assessment for Cancer and Early Detection (GRACE) program uses family and medical history as well as digital mammography, AI and embedded algorithms to assess a patient’s potential risk of developing breast cancer. Wes Walker, M.D., vice president of genomics & precision health at 55-hospital AdventHealth, spoke with Healthcare Innovation about the impact the program is having. 

The Florida-headquartered health system is embedding cancer risk assessment into primary care and OB/GYN workflows. Walker has strategic and operational responsibility for genomics initiatives in AdventHealth’s Central Florida division. But he also works to identify and develop pilots around genomics and precision health that can be scaled system-wide. Walker said health systems are at an inflection point in terms of being able to operationalize precision medicine at scale. 

Healthcare Innovation: With the GRACE program, did you see an opportunity to identify higher risk patients earlier in the process? And was age primarily how people were stratified to be screened before this?

Walker: Yes, we saw in the guidelines and in more advanced clinical practices, as well as some of the accreditation programs, that they were looking for more tailored, personalized, evidence-based risk assessment methodologies other than just age. When we looked at our organization, we found there was an opportunity. We put together a multidisciplinary team to develop a program on both the technical side and the high-touch side.

HCI: How is AI involved and how does a radiologist work with the risk analysis tool?

Walker: We work with a company called Lunit that has an AI module that provides predictions and recommendations for radiologists related to breast density. Breast density is a very important component of the most advanced risk assessment models. 

Here is how the program works: Let’s say somebody schedules their mammogram. Two weeks before they come in, they get a survey to gather information that's not already in Epic that can be used in these risk assessment modules. They come in and get their mammogram, and the Lunit module with the AI enablement provides a recommendation. The radiologist can use that — and I would say over 90% of the time they use that. A risk assessment is calculated. That is provided to the radiologist. All of this is occurring in Epic, and can be seen by the entire healthcare team. If the patient is identified as high-risk, that information automatically flows to the nurse navigator to enable them to reach out and have a conversation around what that high-risk status means. 

The program has two components. There's high-touch and then high-tech. The tech is on the front end, and high-touch is on the back end in terms of making sure that this isn't just a number or a radiology report that goes into the chart, but rather there is a human who has deep expertise in this area. Our nurse navigators, on average, have 15 years of oncology nurse navigator experience. They can provide education but also provide that emotional support, because finding out you're high-risk for breast cancer can be a very stressful experience.

HCI: I saw you quoted as saying that roughly 20% of the women who come in to get a mammogram are actually at high risk. As you developed the program, did you have to add more nurse navigator positions?

Walker: Yes, we did hire a team of nurse navigators, but we've also used the population health tools in Epic to automate the things that can be automated and make them incredibly efficient. The important part of that is having our nurse navigator lead and her team as a part of the design process to optimize the EMR in such a way that it makes them incredibly efficient. That's only going to become more powerful. There are additional automations available. We’re going to be hiring additional nurse navigators as we scale this program. But because of the variety of tools we have, they will become increasingly efficient and productive in terms of the number of patients they're able to to manage and impact.

HCI: Is there a good way to represent the family health history in Epic?

Walker: Yes, there is the pedigree. But you also have to turn that into discrete data that rolls into Lunit’s risk pathway software and is incorporated into creating what we consider the most advanced risk modeling available.

Lunit’s risk pathways module takes the NCCN guidelines, which are really the gold standard cancer guidelines, and they instantiate them into the software. Those guidelines are updated, often twice a year, so there's a lot of coding that has to go on behind the scenes to make sure that those guidelines are up to date.

It serves two purposes. One is to use the best evidence-based methodology to identify patients who are high-risk, and two, if genetic testing is recommended or additional imaging is recommended, there's going to be a pre-authorization process, and that information helps with that pre-authorization process.

HCI: Is there a role for genetic counselors sometimes, too?

Walker: There is. We have some at AdventHealth, but there are just not enough of them. They're very hard to hire, so we also partner with genetic testing companies that have genetic counselors on staff as well.

HCI: You’ve done this impressive work with mammography. Could this also be applied to colorectal cancer screening or in other areas?

Walker: NCCN guidelines also exist in terms of hereditary risk assessment for colon cancer, and they've just updated them as well. In conjunction with Lunit, we've updated our modules. Now we're looking at breast, ovarian, colon, pancreatic, gastric, endometrial, and prostate. 

We're not only going to expand the number of cancers that we're looking at from a hereditary risk assessment standpoint, we're also in the process of expanding out to primary care providers and then OB/GYN. Ultimately our vision is that this risk assessment will be the standard of care for every adult cared for by AdventHealth at some point in the not-too distant future.

HCI: Is it expected that following these high-risk patients more closely and working with them earlier will lead to better outcomes. Will there be a way to measure that?

Walker: Yes. We're in the process of measuring that, but it will take years. We're already tracking all the downstream care that takes place, as well as downstream outcomes, cancer diagnoses, staging, and treatment. We're already working on several manuscripts for publication.

HCI: If we looked across the country, how many health systems have programs like GRACE? 

Walker: We're not aware of many that do. We've heard from Epic that we're a leader in this space. There are systems that are doing good work. We're not aware of another system that has done all the behind-the-scenes integration to integrate the radiology and the PACS and all the work it takes to to truly integrate Lunit risk pathways into Epic, and couple that with the nurse navigator component. 

HCI: Obviously, it's a lot of hard work to do all that integration, as you say. But in another sense, it seems like a no-brainer that you'd want to get ahead of these diagnoses.

Walker: Yes, you would. There are a couple reasons why perhaps it hasn't happened. The guidelines weren't mature enough. Also, for a while there was some pushback on genomic testing from payers, and it was still so new. But the cost of genetic testing has also come down significantly, and will continue to because of the development of new technologies. I see this as an area that will be growing.

HCI: Has there been progress in pharmacogenomics, too?

Walker: There’s also been change there. Pharmacogenomics has been available for almost 15 years. With some of the use cases around cancer chemotherapy, and that getting incorporated into guidelines, and with some Medicare's reimbursement changes, I think you're also going to see a pretty significant uptick there.

HCI: Could you describe the rollout of GRACE? I read it started in 2024.

Walker: Yes. We piloted it in two mammography centers. We did a couple of iterative, improvement cycles. You’ve got to make your mistakes small so that you can get it to something that is scalable. And then we scaled across our the rest of our footprint in Central Florida.

HCI: And that’s 16 centers, right?

Walker: Yes, 16 centers. Then Rocky Mountain went live, and and now it's being implemented across all mammography facilities in the AdventHealth national footprint.