Research Article: Using a machine learning approach to predict outcome after surgery for degenerative cervical myelopathy

Date Published: April 4, 2019

Publisher: Public Library of Science

Author(s): Zamir G. Merali, Christopher D. Witiw, Jetan H. Badhiwala, Jefferson R. Wilson, Michael G. Fehlings, Carmen L.A.M. Vleggeert-Lankamp.

http://doi.org/10.1371/journal.pone.0215133

Abstract

Degenerative cervical myelopathy (DCM) is a spinal cord condition that results in progressive non-traumatic compression of the cervical spinal cord. Spine surgeons must consider a large quantity of information relating to disease presentation, imaging features, and patient characteristics to determine if a patient will benefit from surgery for DCM. We applied a supervised machine learning approach to develop a classification model to predict individual patient outcome after surgery for DCM. Patients undergoing surgery for DCM as a part of the AOSpine CSM-NA or CSM-I prospective, multi-centre studies were included in the analysis. Out of 757 patients 605, 583, and 539 patients had complete follow-up information at 6, 12, and 24 months respectively and were included in the analysis. The primary outcome was improvement in the SF-6D quality of life indicator score by the minimum clinically important difference (MCID). The secondary outcome was improvement in the modified Japanese Orthopedic Association (mJOA) score by the MCID. Predictor variables reflected information about pre-operative disease severity, disease presentation, patient demographics, and comorbidities. A machine learning approach of feature engineering, data pre-processing, and model optimization was used to create the most accurate predictive model of outcome after surgery for DCM. Following data pre-processing 48, 108, and 101 features were chosen for model training at 6, 12, and 24 months respectively. The best performing predictive model used a random forest structure and had an average area under the curve (AUC) of 0.70, classification accuracy of 77%, and sensitivity of 78% when evaluated on a testing cohort that was not used for model training. Worse pre-operative disease severity, longer duration of DCM symptoms, older age, higher body weight, and current smoking status were associated with worse surgical outcomes. We developed a model that predicted positive surgical outcome for DCM with good accuracy at the individual patient level on an independent testing cohort. Our analysis demonstrates the applicability of machine-learning to predictive modeling in spine surgery.

Partial Text

Degenerative cervical myelopathy (DCM) is a spinal cord condition that results in progressive non-traumatic compression of the cervical spinal cord[1,2]. DCM is the most common cause of spinal cord dysfunction globally and can result in significant impairment in quality of life and function among affected patients[3]. Surgical decompression is the preferred treatment to alter the course of DCM and has been shown to improve functional outcome and quality of life in most but not all patients[4]. Indeed, the variability in extent of improvement in patients undergoing surgery for DCM is striking[4–8].

Surgical decompression is the preferred treatment for DCM and can result in long-term improvement of myelopathic symptoms and quality of life in the majority of patients although the extent of improvement can vary widely[4]. In this study we applied a machine learning approach to a multi-centre prospective database and were able to predict outcome after surgery for DCM at the individual patient level with good performance. In addition we identified the following pre-operative variables as important predictors of surgical outcome: older age, duration of DCM symptoms, pre-operative disease severity, body weight, and smoking status. To our knowledge this is the first study to apply a machine learning approach to predict surgical outcome after DCM. These results can be applied to guide surgical decision-making and support the results of previous studies using classical statistical methods.

We retrospectively applied a machine learning approach to a multi-centre cohort of patients who underwent surgical decompression for DCM. Our final random forest model was able to predict positive surgical outcome with good accuracy at the independent patient level on an independent testing cohort that was not used for model training. Our model identified worse pre-operative disease severity, longer duration of DCM symptoms, older age, higher body weight, and current smoking status as being associated with worse surgical outcomes. To our knowledge our model, using a machine learning approach, achieved a higher accuracy than previously published models. We identified longer duration of DCM symptoms, worse pre-operative disease severity, higher age, higher body weight, and current smoking status as being associated with worse surgical outcomes, which supports the results of previous studies. Our analysis demonstrates the applicability of machine-learning to predictive modeling in spine surgery.

 

Source:

http://doi.org/10.1371/journal.pone.0215133

 

Leave a Reply

Your email address will not be published.