Related Experiment Video
Updated: Sep 13, 2025
Diffusion Tensor Magnetic Resonance Imaging in Chronic Spinal Cord Compression
Published on: May 7, 2019
Artificial intelligence and machine learning in the management of patients with degenerative cervical myelopathy: a
Vikas N Vattipally1, Ritvik R Jillala1, Carlos A Aude1
1Department of Neurosurgery, Johns Hopkins University School of Medicine, Baltimore, MD, USA.
Introduction:
Degenerative cervical myelopathy (DCM) is a debilitating condition caused by compression of the spinal cord. Despite established surgical treatments, accurate diagnosis and prognostication remain challenging in part due to the variability in clinical presentation and lack of screening tools. Machine learning (ML) has emerged as a promising approach to address these challenges through its predictive capabilities for diagnosis, decision-making, and prognostication. Given the recent advent of ML, there is a need to systematically synthesize its applications to the treatment of patients with DCM.
Evidence Acquisition:
A systematic review was performed in accordance with PRISMA guidelines. We searched five databases (PubMed, Embase, Cochrane, Scopus, Web of Science) in November 2024 and included studies employing predictive ML techniques among a population of patients with DCM. Studies primarily focused on ML applications to neuroimaging were excluded. Variables such as study focus, number of patients with DCM, and ML approaches used were extracted.
Evidence Synthesis:
Thirty full-text studies were included in this review. These studies encompassed 11,407 patients, with 84% (N.=9615) holding a diagnosis of DCM. Most studies (N.=16, 53%) used ML to predict outcomes for patients with DCM, including functional recovery, quality-of-life, and postoperative complications. Thirteen studies (43%) focused on the diagnosis of DCM using ML-augmented screening tools, and the remaining study focused on surgical decision-making. Support vector machine was the most used ML approach (N.=14 studies, 47%) followed by random forest (N.=8 studies, 27%). Throughout the studies included, ML algorithm predictions were demonstrated to outperform traditional statistical methods.
Conclusions:
ML models are a promising step forward for diagnosis, clinical decision-making, and prognostication for patients with DCM. Further validation in large, multi-institutional cohorts is needed to help improve translatability to clinical practice.
