EMG analysis of patients with preliminary diagnosis of carpal tunnel syndrome using artificial intelligence

Objective: Carpal tunnel syndrome (CTS) is the most common compression neuropathy, with electromyography (EMG) remaining the gold standard for diagnosis. Advances in artificial intelligence (AI) have revolutionized healthcare, enhancing disease classification, diagnosis, treatment, and prognosis. AI models have shown high accuracy in analyzing EMG signals. This study aims to evaluate EMG parameters using AI algorithms to advance their application in healthcare. Materials and Methods: The data used in this study were collected from the Electroneurophysiology Laboratory of the Neurology Department of Internal Medicine at our hospital between October 1, 2011, and June 30, 2023. A total of 3,298 EMG records from 1,649 patients (one for each hand) were included in the study. The collected EMG data were preprocessed and analyzed using machine learning algorithms. The machine learning algorithms employed in this study include "Decision Tree," "K-Nearest Neighbour," "Support Vector Machine," "Random Forest," "Naive Bayes," "CatBoost," "XGBoost," and "Light GBM." Results:Although the accuracy rates of all models were high, CatBoost, Random Forest, and XGBoost achieved the highest accuracy rates (90%, 90%, and 89%, respectively). KNN and Naive Bayes performed comparatively lower in terms of accuracy (77% and 79%, respectively). Conclusion: The machine learning methods used in this thesis generally demonstrated high success in predicting CTS diagnosis and severity. However, the study highlights some limitations in predicting the "mild CTS" and "severe CTS" groups Keywords: Deep Learning, Electromyography, Carpal Tunnel Syndrome, Machine Learning, Artificial Intelligence