Machine learning shows potential to prevent costly, ineffective back surgery

Editor's Note

Although lumbar disc herniation surgery can alleviate symptoms of back pain and even prevent disability, the costs exceed the benefits for a sizeable minority of patients, researchers write in a study published February 7 in Jama Network. To address this concern, they investigated whether machine learning algorithms that predict individual prognoses can influence decision-making, help manage patients’ expectations prior to the surgery, and reduce the cost burden on health systems from ineffective and costly care.

“Machine learning algorithms are powerful tools for analyzing large amounts of data and have gained traction in recent years, but their use for outcome prediction in spine surgery remains nascent,” the researchers write. However, data available in national spine registries provide ample opportunity to leverage new tools. Specifically, they aimed "to develop and validate machine learning models for predicting improvement in disability and pain 12 months after lumbar disc herniation surgery.”

The prospective, multicenter, registry-based prognostic study was conducted on a cohort of individuals undergoing lumbar disc herniation surgery from January 1, 2007, to May 31, 2021. Patients in the Norwegian Registry for Spine Surgery from all public and private hospitals in Norway performing spine surgery were included. Data analysis was performed from January to June 2023.

In predicting treatment success 12 months after surgery, “internal-external cross-validation showed that our models had consistently good calibration when applied to the different geographic regions and private hospitals, and good discrimination with C statistics 0.81 or greater for disability and 0.74 or greater for pain,” researchers write. “The models integrated 25 routinely available preoperative features and should be readily implementable in standard clinical settings at the point of surgical decision-making. The discriminative performance of our models is generally similar to or better than earlier prediction models for disability and pain improvement following lumbar discectomy.”