Baptist Memorial Health Care Corp. serves as a vital lifeline across the Mid-South, with more than 20 hospitals, 100 imaging locations, and more than 3,000 staffed beds across Tennessee, Mississippi and Arkansas.

THE CHALLENGE

The organization long has understood a responsibility in confronting the region’s disproportionate burden of lung cancer. In the Mid-South, lung cancer mortality rates are nearly double the national average.

In 2020, it made a public commitment to reduce lung cancer deaths by 25% by 2030 – what it calls the "Mid-South Miracle." Achieving that goal required the health system to strengthen both its lung cancer screening program and approach to incidental pulmonary nodules.

Processes relied heavily on manual work. Teams were pulling information from multiple systems, validating patients, re-validating them, and working through lists again and again just to make sure no one was missed.

Staff were using three, four, sometimes five different systems, and none of them worked well enough for the mission the organization was trying to accomplish. The effort required created frustration and inefficiency, and in some cases contributed to staff turnover. As the footprint expanded across three states, these challenges only intensified.

"Every hospital had its own process, and because of that, there was no consistent way to ensure patients were identified and moved into the right pathways," said Parker Harris, CEO of Baptist Memorial Hospital-Tipton and system oncology service line administrator. "Our teams were doing double work – sometimes triple work – trying to keep up with the volume, and we knew it wasn’t sustainable.

"The more we grew as a health system, the clearer it became we needed a reliable, systematic way to identify and manage patients across our entire footprint," he continued. "Without that, we couldn’t guarantee timely follow-up, which is essential for early detection and better outcomes."

PROPOSAL

Baptist Memorial knew it needed to make a change and knew its manual processes would not be a long-term solution for reducing lung cancer mortality across the Mid-South.

"Around that time, we were being inundated with artificial intelligence opportunities, so we wanted to evaluate our options carefully and strategically," Harris recalled. "We received an internal grant from our Baptist Foundation to conduct a formal evaluation, and we partnered with an outside consulting firm to help us determine which oncology challenges we should prioritize for AI and which vendors were best positioned to support those needs.

"Among the many AI opportunities we evaluated, the most strategic and foundational one for us was the ability to identify incidental findings consistently and at scale across every facility in our health system," he added. "As a community-based cancer center, our mission is to ensure that a patient in Booneville, Mississippi, has the same opportunity for timely care as a patient in Memphis, Tennessee, or Jonesboro, Arkansas."

Accurate, systemwide identification of incidental findings was essential to providing that level of equitable care, he said.

The most compelling data point that moved incidental findings to the top of the priority list was the prevalence study the organization conducted with Eon, vendor of an AI-powered longitudinal care management platform.

"We believed our prevalence rate of incidental pulmonary nodules was 0.362%," Harris noted. "Eon was confident it was significantly higher. I asked them to prove it and provided a test volume of 2,000 radiology reports.

"Their computational linguistics model analyzed the dictated text, and the prevalence came back at 2.28%," he continued. "That difference represented a substantial number of patients who were at risk of being overlooked. The study helped our leadership team quantify the scale of the opportunity to intervene earlier for patients who otherwise might not have been brought into follow-up."

But accurate identification alone was not enough for a health system the size of Baptist Memorial. One of the biggest parts of the vendor’s proposal – and one of the main reasons staff selected the company – was its ability to help the health system absorb the volume that would come with better detection.

"We knew implementing an identification engine without the capacity to manage the patients it surfaced would not be sustainable," Harris explained. "What stood out was the combination of precise identification and the additional support available to help manage and navigate the patients who needed follow-up.

"That dual capability – accurate detection paired with the operational resources required to ensure patients entered the correct pathway – addressed both sides of our challenge," he continued. "It offered a clear path to reduce the manual burden on our teams, standardize processes and support the scale of our mission."

MEETING THE CHALLENGE

Baptist Memorial uses Epic for its electronic health record across the health system, and Eon is fully integrated with Epic. Every radiologist-dictated report is delivered to Eon over an HL7 feed as soon as the report is signed.

Eon’s computational linguistics engine analyzes each report immediately. When an incidental pulmonary nodule is identified, the finding and relevant clinical characteristics are extracted and displayed in the system.

"Through our FHIR integration with Epic, additional patient risk factor data – such as smoking history, prior cancer history, primary care provider and preferred language – automatically is pulled in, allowing each case to be reviewed with a complete and current clinical profile," Harris said. "Eon’s Navigation Services function as an extension of our own navigation teams, supporting the operational work of case review, pathway activation and patient coordination for the large number of findings that surface each day.

"The vendor’s Navigation Services team manages the platform and reviews each case," he continued. "When the next step is clear based on established clinical guidelines and Baptist-approved protocols, the technology automatically generates the recommended follow-up," he added.

This ensures the next step is driven by guidelines and Baptist Memorial protocols – not solely by what a radiologist may or may not document in the dictated report. This integration has allowed staff to establish a consistent workflow across all hospitals and imaging locations in the health system.

"All communication – including provider notifications, patient outreach and navigator interactions – is documented back into the Epic chart as documented encounters and is also recorded within Eon," Harris noted.

"When the guidelines or next steps are not clear, those high-risk or ambiguous cases are routed to the local Baptist navigation teams and prioritized within a dedicated worklist for additional review," he continued. "This ensures more complex cases receive timely clinical oversight by providers or a multidisciplinary team."

Eon also automatically captures unexpected diagnostic imaging that relates to the incidental finding. When new imaging shows progression or changes the clinical picture, the care plan dynamically updates so patients do not remain on an outdated follow-up pathway.

"This approach also enabled us to manage a significant increase in volume," he explained. "We more than doubled the number of pulmonary nodules identified in the first year, and the Navigation Services helped absorb and manage that growth so our local teams could focus on patient engagement and clinical decision making.

"This allowed us to scale the program without adding staff while ensuring patients received timely, coordinated follow-up based on the most current clinical information," he added.

RESULTS

One of the most meaningful outcomes for Baptist Memorial has been the increase in incidental pulmonary nodules identified. In 2024, it identified 7,160 incidental findings. In 2025, it is on track to identify more than 18,000.

"53% of these nodules are more than 8mm, which means they represent a higher-risk group that needs timely clinical review," Harris reported. "For these high-risk incidental pulmonary nodules, the time from when the radiologist submits the dictated report to when the case is sent to a provider for review now averages about two days so we are able to get high-risk patients identified and into care much faster.

"But identifying patients is only the first step," he noted. "Identification doesn’t matter if patients do not come back for follow-up. Historically, the Mid-South has had some of the lowest lung cancer screening rates in the country, which makes our follow-up performance especially meaningful."

89% of patients with incidental pulmonary nodules returned for their recommended follow-up, and 77% of lung cancer screening patients returned as well. Sustained follow-up is critical for early detection and has been one of the strongest indicators of program success.

"To date, we have identified 625 lung cancers, directly attributable to cases surfaced and managed through the vendor platform," he said. "These are individuals whose cancers were detected and addressed because the findings were identified accurately and the follow-up was coordinated effectively.

"From a financial perspective, while patient outcomes remain the most important metric, we also saw $3.44 million in incremental direct margin over 12 months from downstream PET/CT, interventional radiology, interventional pulmonology and thoracic surgery services," he continued.

The health system experienced significant increases in annual volumes, including a 60% increase in interventional pulmonary procedures, 99% increase in interventional radiology procedures, 65% increase in PET/CT volume and an increase in thoracic surgeries. These results contributed to the organization achieving a return on investment in the Eon system in seven months.

*Follow Bill’s health IT coverage on LinkedIn: *Bill Siwicki *Email him: *[email protected] Healthcare IT News is a HIMSS Media publication.

WATCH NOW: Managing government policy and regulation in healthcare AI adoption