Related Experiment Video
Updated: Sep 28, 2025
05:19
Author Spotlight: Therapeutic Benefit of Closed-Loop Deep Brain Stimulation in Depression Treatment
Published on: July 7, 2023
2.6K
Machine Learning to Predict Successful Opioid Dose Reduction or Stabilization After Spinal Cord Stimulation
Syed M Adil1, Lefko T Charalambous1, Shashank Rajkumar1
1Department of Neurosurgery, Duke University Medical Center, Durham, North Carolina, USA.
Neurosurgery
|April 6, 2022
Summary
Predicting spinal cord stimulation (SCS) success in reducing opioid use is now possible with machine learning. A simplified logistic regression model shows comparable performance to deep learning, aiding clinical decisions.
Area of Science:
- Pain Management
- Neurosurgery
- Health Informatics
Background:
- Spinal cord stimulation (SCS) effectively reduces opioid usage in select patients.
- Currently, no objective preoperative measure predicts SCS success for opioid reduction.
Purpose of the Study:
- Develop machine learning models to predict successful opioid reduction or stabilization after SCS.
- Compare deep learning (DNN) performance against logistic regression (LR).
Main Methods:
- Utilized IBM MarketScan data (2010-2015) for 7022 patients undergoing SCS.
- Developed LR and DNN models using 30 predictors, focusing on medication patterns and comorbidities.
- Assessed model performance using nested 5-fold cross-validation and AUROC.
Main Results:
- 66.9% of patients achieved successful surgery (opioid stability/reduction at 1 year).
- A 5-variable LR model performed comparably to the full 30-variable version (AUROC difference <0.01).
- DNN and simplified LR models showed similar AUROCs (0.740 vs. 0.737).
Conclusions:
- Presents the first machine learning models for predicting SCS success in opioid reduction.
- Simplified LR model demonstrates comparable performance to DNN and highlights pharmacologic patterns.
- This interpretable LR model can aid patient and surgeon decision-making for SCS.
