How AI is impacting care


Artificial Intelligence-powered software has quickly become mainstream as leaders across many industries weigh the limitations and vast possibilities of these tools.

Commonly known as AI, the technology refers to the development of computer systems that can simulate human thinking and problem-solving capacities.

In the health care sector, experts have spent years researching the ability to harness these tools to improve patient care and the way health care offices function, from billing to scheduling.

Cancer imaging is an early area of adoption where researchers have explored the use of AI to improve work efficiency and enhance diagnostic performance.

While AI-powered technology has been available for many years, over the last three years the capabilities of computers have grown tenfold, says Sotirios Stathakis, the Dr. Charles M. Smith chief of physics at Mary Bird Perkins Cancer Center.

“That’s why we have seen this explosion in AI,” he says, “and there is more to come.”

One AI-based tool his team will soon be using is the FDA-approved MR-Box by TheraPanacea. The software uses AI technology to create synthetic CT scans using information from MRI scans. While the MRI scan is used to accurately identify the location of cancer, the CT scan is used to calculate the dose for radiotherapy treatment. CT scans require an additional visit from the patient.

“With this software we are cutting down one whole appointment for the patient, so less time at the center. We are still doing the MRI but using this tool to create the CT scan,” Stathakis says. “It’s based on information from the MRI and the scan is identical to one we would have gotten if the person had gone for a CT scan.”

Additionally, his team is also using the FDA-approved software TheraPanacea Annotate to accurately contour and identify organs in the body. Stathakis explains that a challenge in radiation therapy is ensuring that normal tissue is not being harmed when killing tumor cells. This technology creates a more accurate image of where organs are located. It’s work that would have once taken hours but now takes minutes, he says.

(Courtesy Mary Bird Perkins Cancer Center)

AI AND PATIENT CARE

Experts say that AI-powered tools have also improved the way that patients interact with the health care system. Beyond imaging, tools using AI technology have been implemented at Mary Bird Perkins to help with scheduling and to identify patients who could benefit from clinical trials.

The center implemented a software called OncoTrials/AI in 2022, which screens new and continuing patient files daily to see whether they can be referred for a clinical trial.

Mandy Shipp, director of clinical research, says that previously staff members only had the capacity to screen newly diagnosed patients. Now staff members simply screen the charts of individuals identified by OncoTrials as potential participants for a trial.

Prior to implementing this tool in 2021, a total of 5,505 patients were screened for clinical trials. In 2023, the first full year of using the software, that number jumped to 12,328.

Additionally, the tool allows Shipp’s team to assess whether a trial is a good fit for the patient population. The team can build a query of certain characteristics and diagnoses that a specific trial is looking for to see whether there is a match, she says.

“So basically, we can assess if (a trial) will help the patients being seen at our center,” she says.

Prior to using this tool, patients were being screened and enrolled into approximately 34% of trials opening at the site. Now the team is screening and enrolling patients in about 85% of the trials.

Shipp says the software allows her team to use its resources more efficiently and has vastly improved its ability to identify patients who can benefit from trials.

AI-powered tools have also been used to improve patient scheduling.

“At infusion centers and our oncology offices we have peak times of the day,” says Brandy Knight, senior director of infusion services at Mary Bird Perkins. “It can start out slow but then get really busy. During those busy times, it can be very chaotic, with patients waiting for labs or to see the doctor to get their infusions.”

In an effort to decrease scheduling issues, they began using a system called OncoSmart, by Smirta, in January. The system recommends optimal times to schedule patients, thus reducing wait times and allowing staff members more time with patients.

PREDICTIVE ANALYTICS

At Our Lady of the Lake Regional Medical Center, AI tools using predictive analytics help monitor and catch certain diseases. Information from a patient’s chart is analyzed to predict if they are deteriorating.

“It gives our nurses an idea of the overall status of a patient’s   well-being and also helps in continuity of care during a shift change,” says Dr. Tonya Jagneaux, chief medical informatics officer.

Another AI tool is used to help predict severe infections, though it’s not always accurate.

“It’s often accurate after the fact and not before. We have been battling with tools that are not always 100 percent accurate with prediction and sometimes they are more functional after the fact,” Jagneaux adds.

She recently collaborated with LSU Provost Roy Haggerty, consultant Henry Hays and deep learning expert James Ghawaly to put together a class focusing on AI development and deployment for business. Working with a team of four honors medical students, they are working to develop AI tools that address specific health care challenges.

Last fall, they developed a tool that provides health care information on a mobile device for patients with heart disease. The software is aimed at educating patients about their disease and connecting them with resources in their communities.

The students are now working on a tool that can help predict the likelihood of a patient developing sepsis—a life-threatening condition—based on their recorded symptoms and history.

Jagneaux again stresses that predictive technology is not always accurate, though there are times when it is spot on.

TAKING NOTES: Suki AI is being used by 30 providers in the Our Lady of the Lake Regional Medical Center system. With the patient’s consent, it is used to generate notes from conversations between the patient and their provider. The visit summary is then reviewed by the provider. (iStock)

“We battle noise—there is harm in saying something is wrong too many times—you want to grab information that might save a life, but you don’t want to give procedures to someone where it is not necessary,” she says. “You have to pause and think what the right thing for our patients is.”

Also in the last year, 30 providers within the Our Lady of the Lake system began using Suki AI software. With the patient’s consent, Suki Ai is used to generate relevant notes from conversations between the patient and their provider. The visit summary is then reviewed by the provider.

Suki AI has been used extensively in clinics and ambulatory spaces, as well as in internal and family medicine where there are not a lot of chronic conditions, says Will Landry, senior vice president and chief information officer at Franciscan Missionaries of Our Lady Health System.

Although AI-powered tools present opportunities to address a wide variety of challenges in the health care sector, Landry says that because the technology is rapidly evolving, it’s important to carefully asses the value when choosing what to implement.

“Internally we call it augmented intelligence,” he says. “We actively want these tools to assist our staff. We don’t want to use it for decision-making. The provider and patient relationship has to be at the center of the experience.”