Data proves its value in everything from predicting surgical outcomes to securing PPE
More hospital systems are turning to health data analytics—not only as they struggle to cope with and plan for a pandemic, but also to gain valuable insights into overall operations and patient outcomes. Data analytics can use machine learning and artificial intelligence to glean information about everything from resource utilization to effective treatments and disease trajectories.
“Analytics is a big topic for us,” says Shay Bess, M.D., an orthopedic spine surgeon at Denver International Spine Center and President of the International Spine Study Group, the nonprofit research foundation dedicated to advancing treatments for people with spine deformities—and one of the most productive spine study groups in the world. Dr. Bess, a HealthTrust Physician Advisor, also works with the organization on some of its clinical data analytics initiatives.
Anticipating patient outcomes
“It is an extraordinarily exciting time—to be able to use data not only to evaluate how patients are doing, but to predict how the patient is going to do,” Dr. Bess says.
“We’re learning more and more that we can predict how patients respond to nonoperative and operative approaches. And we can also predict what their responses will be on a health-related outcome measure in four to five years based on the intervention.”
For instance, Dr. Bess notes, clinicians typically categorize patients with scoliosis as having “large” or “small” scoliosis. But analyzing data points can take it to another level.
“The computer may see them as highly disabled or not disabled at all,” Dr. Bess explains. The magnitude of their disease and its impact on their quality of life can determine treatment options and outcomes.
Using the patient’s genotype, researchers can even predict whether patients are likely to develop an opioid addiction, while tissue and blood samples provide data to assess the patient’s physiologic age, a better predictor of outcomes than chronological age.
Dr. Bess notes that one study compared how a surgeon predicted a patient’s prognosis versus how the computer thought the patient would do. “The computer was much more accurate,” he says.
“Analytics has turned upside down how we think about patients,” says Dr. Bess. “It also shows us that the rudimentary criteria using rapidly outdated data is not going to work. There is a better way to evaluate patients.”
Data during a pandemic
During this time of COVID-19, the gush of data and information at the local, state, national and global levels is overwhelming. Using that data effectively is integral to success.
HealthTrust is at the forefront of that effort, working with Dr. Bess to use analytics to estimate consumption rates for personal protective equipment (PPE), medications, and ICU and ventilator use for COVID-19 patients. The goal, Dr. Bess explains, is to predict consumption for individual hospitals based on volume and size so they can prepare for a surge in patients.
“If the algorithm shows that a specific hospital will have enough equipment, it may prevent a full shutdown of elective surgeries and other behavior changes,” he says. That would be critical for hospitals, which are experiencing severe financial strain due to canceled elective surgeries and procedures during the pandemic.
However, Dr. Bess stresses that the human element is still critical in analyzing data. “Predictive analytics might get you from A to B quickly, but that might not be the right direction,” he says. “It takes people collaborating to come up with the right questions to feed into the analytics program, which can then take us to the answer in a very quick and accurate fashion.”Share Email