Artificial intelligence (AI) can be used to bring efficiency and automation to the decision-making process of determining surgical candidates for patients with lumbar spinal stenosis (LSS), with performance comparable to multidisciplinary physician teams . this is, European Spine Journal Raphael Mourad (Remedy Logic, New York, USA) others.
The study also found that imaging in combination with specific clinical variables such as motor impairment and pain were important predictors of surgical candidates.
According to the research team, “Given that physicians and other health care providers must obtain pre-approval from health insurance companies before certain services can be provided, our model is fast and efficient at limited cost. It can serve as an important tool for making efficient decisions.”
Researchers developed a new hybrid AI model that calculates the likelihood of recommending spine surgery for LSS based on patient demographic factors, clinical manifestations, and magnetic resonance imaging (MRI) findings. I suggested.
The hybrid model is a random forest model trained from medical vignette data reviewed by surgeons and peer-reviewed literature and expert opinion from multidisciplinary teams in spine surgery, rehabilitation medicine, interventional and diagnostic radiology. It combines expert Bayesian network models built from Sets of 400 and 100 medical vignettes reviewed by surgeons were used for training and testing.
An independent committee of five spine surgeons (fellowship-trained spine surgeons with at least five years of practical experience) was established. A panel reviewed 500 medical vignettes to determine the probability of surgical recommendation for each vignette.
They found that the model showed high prediction accuracy, with a root mean square error (RMSE) of 0.0964 between model predictions and the ground truth, whereas the average RMSE between individual physician recommendations and the ground truth was I found it to be 0.1940.
For dichotomous classification, the area under receiver operating characteristic (AUROC) and Cohen’s kappa were 0.9266 and 0.6298, whereas the corresponding average metrics based on individual physician recommendations were 0.8412 and 0.5659, respectively.
talk Spinal News InternationalRemedy Logic CEO Andrej Rusakov said: We plan to package the knowledge of America’s leading orthopedic and neurosurgeons into AI and make it available at low cost to anyone with an internet connection. We are committed to making independent, unbiased and highly accurate medical advice available to everyone around the world. “
