Alexis is director of content at AIMed, with responsibility for the research, development and delivery of products across events, digital and publishing. A highly experienced events executive with a career focus on the intersection between healthcare and technology, he is also a school governor leading on teaching, learning, and quality of education.
The University of Virginia Center for Advanced Medical Analytics is committed to improving patient outcomes by getting hospitals ready to deploy AI solutions…
“Every day in healthcare, we react to something that happens to patients. A lot of times, they are moving steadily down the track, hours before a formal diagnosis occurs,” says Dr. Randall Moorman, Clinical Cardiologist and Founding Director of the University of Virginia Center for Advanced Medical Analytics. “Patterns in data presented to clinicians are ‘brakes’ to apply before patients head towards a cliff and thereby prevent unnecessary harm and increase costs of care.”
Dr. Moorman is an internationally recognized pioneer in the field of predictive analytics monitoring. Twenty years ago, he and his colleagues created a visual risk display called HeRO, to alert clinicians, to the chances of premature babies exhibiting abnormal heart rate patterns, developing sepsis. In a large, randomized trial, he and colleagues discovered that 3000 at-risk and low birth weight babies born in nine different hospitals who had HeRO displays at their bedsides were 20% less likely to die.
Dr. Moorman believes by uncovering and connecting the millions of data points and presenting them using simple visualization, physicians will begin to see “signatures” of potential series events before clinical signs emerge. He cites a large flat screen monitor at the nursing station of an intensive care unit (ICU). Here, each bed number is shown on the monitor with a simple visualization that looks like a comet, rising or falling, depending on the acuity of the patient.
The head of the comet signals a patient’s risk of entering clinical adversity in the next six to 12 hours. The length of the comet tail and brightness tell how rapidly the patient has deteriorated or progressed in the past three hours. “These visualizations are straightforward, intuitive, and immediately actionable,” Dr. Moorman remarks. “The comet goes up when a patient is at risk and goes down when the risk decreases. In fact, a recent publication demonstrated the use of such method reduced the diagnosis of sepsis shock by 52%.”
In fact, ‘CoMet’ refers to Continuous Monitoring of Event Trajectories. It was an AI program designed by Dr. Moorman which combines real-time, non-stopping data collected from bedside monitoring networks and electronic health records to assess patient’s health when they were in the ICU and tells clinicians if their treatment methods are working. The program was recently deployed to monitor high-risk coronavirus patients, helping clinicians to take a proactive role in attending patients before their condition worsened.
“Vital sign measurements and labs can come too late, but early detection through predictive analytics has the power to improve patients’ outcomes, especially for catastrophic illnesses like COVID-19,” says Dr. Moorman. But he’s not planning to stop there. His colleague, Dr. David J. Stone, affiliate of the Departments of Anesthesiology and Neurosurgery, is presently interrogating some of the scoring systems used in ICU to predict the risk of death.
What Dr. Stone and colleagues found, as detailed in a study published in April, was the scores demonstrated a degree of ethnic bias in their standardized mortality-ratio calibrations with a consistent pattern of mortality overprediction for African American and Hispanic patients as compared to White and Asian patients. This signified that some patients could have been unfairly denied appropriate access to ICU beds, ventilators, and other resources.
“Our major take-home point is that these supposedly objective tools should not be used to make clinical care or triage decisions for individual patients,” says Dr. Stone. “It’s time to develop scoring systems that are more precise than the current one-size-fits-all systems. Incorporating precision socioeconomic and geographical parameters, along with asset of specific biomarkers for a given disease, into future prediction models might take such models less biased and more robust.”
Apart from data, Dr. Stone is also calling on hospitals to create clinical departments committed to developing, researching, and deploying AI to harness its power in transforming patient care. AI shows potential in benefitting healthcare delivery, but it’s often blunted by inconsistent implementations. Thus, Dr. David Stone and colleagues at UVA Health and several other major medical centers, including Dr. Leo Anthony Celi, outlined a plan to make hospitals “AI ready” last summer.
According to the authors, an AI clinical department will drive medical facilities to adopt long-term plans and goals, rather than focusing on disconnected, short-term projects to fulfill immediate needs. The department will connect expertise from different fields, gather resources and support to drive research effort, AI deployment, designing safety metrics that monitor algorithms’ performance over time, and fulfill regulatory requirements.
Although Dr. Stone notes that today’s clinicians may view the idea of a clinical AI department as “frankly, crazy” but he sees the likelihood of AI becoming an essential part of clinical processes in the future. “This is an opportunity to do it right from very near to the beginning of clinical AI’s use, rather than having to repair and replace a flawed system in the future.
“These initiatives should lead to the development of models that will directly benefit the health of our patients, pioneer research that advance the field of clinical AI, focus on its integration into clinical workflows and foster educational programs and fellowships to ensure we are training current practitioners as well as the next generation of leaders in this field.”