HCA Uses AI to Decrease Time From Cancer Diagnosis to First Treatment by Almost a Week

The machine learning models have also helped increase patient retention by more than 50% in a very competitive market, the health system says.

Franklin, TN, February, 2023 /HealthcareITNews/ — HCA Healthcare has 14 hospitals covering a radius of more than 300 miles. The staff members knew they had a large catchment area. However, caseloads were based on registry data that was delayed, and not a true reflection of the provider organization's oncology program as a whole.

THE PROBLEM

"With multiple points of entry, we relied on manual searches through pathology, physician schedules and referrals to identify newly diagnosed cancer cases and patients," said Kristina Rua, RN, an oncology nurse, oncology nurse navigator and president of the Florida Academy of Oncology Nurse and Patient Navigators, who previously was director of oncology navigation services in the East Florida division of HCA Healthcare.

"Not only was this a time-consuming task, but it meant resources were being allocated to the identification of patients versus the actual care of them," she explained. "Nurses no longer were actively involved directly in patient care due to the administrative duties they were tasked with in order to identify newly diagnosed patients."

Delays in identification were incurred as the process began only after a result was available in the patient record or provided to the physician, meaning there could be several days delay in the care team being made aware of the actual diagnosis.

"For patients at high risk for cancer, finding a technology such as an AI tool can put them through treatment in a timely manner, as the AI helps to identify positive diagnoses early on and increases chances of saving lives." - Kristina Rua, RN, Florida Academy of Oncology Nurse and Patient Navigators

"Timeliness to care was affected, as often the awareness of a newly diagnosed patient was several weeks post procedure," she said. "We also knew patients were out-migrating to other institutions. However, we could not quantify this leakage due to lack of data. By alerting a navigator of patients as soon as they were diagnosed, they were able to support the patient very early in their cancer journey, and in return retain them for treatment," she continued. "Lastly, referral patterns were not easily tracked, which ultimately affected our service line growth and development."

PROPOSAL

HCA Healthcare chose Azra AI technology to help. The vendor offered multiple technologies to help solve the complex problems, beginning with early identification of cancer patients through the use of artificial intelligence to surface positive cancer patients from pathological reports. The vendor also offered the ability to surface incidental findings from radiology image text.

"They also offered the option to abstract said data, facilitating the role of cancer registry," Rua explained. "They offered additional efficiencies for the registry by populating more than 50 additional required fields for them and either integrating with the registry software or putting the information in the required format to be uploaded. For our nurse navigator team, there was the option of a documenting system that could track the patient from diagnosis through the cancer continuum with the ability to run reports that would aid administration to grow the navigation program and provide concrete data that would help in service line programmatic growth," she added.

Azra AI offered the ability to centralize the HCA division's data into essentially one platform from which staff could derive data to set service line goals and growth objectives with real-time data and volumes.

MEETING THE CHALLENGE

The program was rolled out division-wide, and focused on identification of pathological reports that indicated a possible cancer diagnosis. These reports were then broken down by facility and tumor site and assigned to the nurse responsible for navigating that disease state.

"The beauty of this process was that now the pathology was hitting the navigator's queue live," Rua noted. "And the technology was integrated into our Meditech EHR. In addition, with the new concurrent abstracting requirements for cancer registry it allowed HCA Healthcare as a whole to centralize its registry process, with less staffing, as there is a certified tumor registrar shortage nationwide," she continued. "We were able to start a case finding file as soon as a patient was diagnosed and pre-populate several data fields for them, in which they were manually reviewing reports for prior."

Azra AI generated a file in the required format (NAACCR) for cancer registry to place cancer case data into Metriq.

RESULTS

"We were able to decrease our time from diagnosis to first treatment by six days," Rua reported. "By alerting a navigator of a patient as soon as they are diagnosed, they are able to step in right away and retain the patient for treatment, while ensuring timely treatment is achieved. Patient retention increased by more than 50%. We are part of a very competitive market," she continued. "Staff efficiencies were achieved, and proven to spend 35% more time with patients post implementing this technology, increasing the navigator's caseload by 71%."

ADVICE FOR OTHERS

AI is making cancer care a lot more efficient, Rua stated. "With the amount of growth in AI and cancer, we'll start to see it be used in precision medicine much more in the future, so we need to prepare, prioritize and be comfortable with this direction now," she advised. "I would tell other healthcare providers to enhance their learning on how genetics affect cancer outcomes. This is an efficient prevention tool and patients often have genetic testing performed to identify potential cases of cancer diagnosis. For patients at high risk for cancer, finding a technology such as an AI tool can put them through treatment in a timely manner, as the AI helps to identify positive diagnoses early on, and increases chances of saving lives," she concluded.

ABOUT AZRA AI

Azra AI uses AI/ML to analyze pathology and radiology reports in near-real time, automating the process by which patients with incidental findings are detected and followed up with for treatment. Health systems’ patient navigators are able to more quickly engage and re-admit patients for further testing and care.