Related Experiment Video
Updated: Sep 21, 2025
09:52
Measuring and Manipulating Functionally Specific Neural Pathways in the Human Motor System with Transcranial Magnetic Stimulation
Published on: February 23, 2020
9.3K
Machine Learning Application of Transcranial Motor-Evoked Potential to Predict Positive Functional Outcomes of
Mohd Redzuan Jamaludin1, Khin Wee Lai1, Joon Huang Chuah2
1Department of Biomedical Engineering, Faculty of Engineering, Universiti Malaya, Kuala Lumpur 50603, Malaysia.
Computational Intelligence and Neuroscience
|May 31, 2022
Summary
Machine learning applied to Transcranial Motor-Evoked Potential (TcMEP) signals shows promise in predicting positive outcomes for lumbar spine surgery patients. This approach may offer a more adaptable alternative to traditional criteria for monitoring nerve integrity.
Area of Science:
- Neurosurgery
- Biomedical Engineering
- Machine Learning
Background:
- Intraoperative neuromonitoring (IONM) is crucial for assessing nervous system integrity during spine surgery.
- Transcranial Motor-Evoked Potentials (TcMEP) are increasingly used in lower lumbar surgery to prevent nerve root injuries and predict functional outcomes.
- Studies indicate a significant correlation between TcMEP signal improvement and positive patient outcomes.
Purpose of the Study:
- To explore the application of machine learning to TcMEP signals for predicting positive functional outcomes in patients undergoing lumbar surgery.
- To evaluate the efficacy of machine learning models compared to existing TcMEP improvement criteria.
Main Methods:
- A machine learning approach was applied to TcMEP signals from 55 patients undergoing lumbar surgery.
- Data were split into 70:30 and 80:20 ratios for training and testing machine learning models.
- Performance was assessed using sensitivity and specificity, comparing machine learning models with established TcMEP improvement criteria.
Main Results:
- The Fine KNN model, using an 80:20 data split, achieved the highest sensitivity (87.5%) and specificity (33.33%).
- The 50% TcMEP improvement criteria demonstrated 83.33% sensitivity and 75% specificity.
- Traditional threshold methods showed decreased reliability and performance variability across different datasets.
Conclusions:
- Machine learning offers a promising and adaptable method for predicting functional outcomes based on TcMEP signals in lumbar spine surgery.
- Further advancements are possible with larger datasets and the incorporation of diverse signal features.
- The study suggests machine learning may provide a more robust alternative to rigid threshold-based criteria for intraoperative monitoring.
