سامانه همایش‌ها TSTA | ثبت‌نام و ارسال مقاله کنفرانس‌ها و کنگره‌های ملی

Home / سی و دومین کنفرانس ملی و دهمین کنفرانس بین المللی مهندسی زیست پزشکی ایران

Improving Effectivity of repetitive Transcranial Magnetic Stimulation in Treatment of Amyotrophic Lateral Sclerosis by Designing New Protocol and Using Machine Learning

Authors :

Ali Abedi¹ Gholamreza Moradi² Reza Sarraf Shirazi³ Mehran Jahed⁴

1- دانشگاه صنعتی امیرکبیر(پلی تکنیک تهران) 2- دانشگاه صنعتی امیرکبیر(پلی تکنیک تهران) 3- دانشگاه صنعتی امیرکبیر(پلی تکنیک تهران) 4- دانشگاه صنعتی شریف، تهران، ایران

Keywords :

repetitive Transcranial Magnetic Stimulation (rTMS)،Machine Learning،Support Vector Machine (SVM)،Amyotrophic Lateral Sclerosis (ALS)،Electroencephalography (EEG)

Abstract :

Repetitive transcranial magnetic stimulation (rTMS) is an effective and old technique of neuromodulation of neuropsychiatric diseases, however patient responses are variable. Finding effective biomarkers that can predict the response to treatment is a critical step in maximizing the therapeutic efficacy. EEG-based features, when integrated with machine learning, provide a promising strategy to the analyze of response. In this work, we explore the effectiveness of EEG-derived features in identifying rTMS responders and non-responders by means of a Support Vector Machine (SVM) model. This study involved 34 ALS patients recruited from a neurology clinic, divided into two groups: 18 received the new rTMS protocol (NP) and 16 followed the Old protocol (OP). Resting-state EEG was acquired in patients before rTMS. Extracted features by using signal processing methods were: time domain (mean amplitude, variance), frequency domain (band power, peak alpha frequency), nonlinear tests (Hjorth parameters, fractal dimension, Hurst exponent). These features were input into SVM classifier. classification performance of SVM model is high, with overall accuracy of 97.3% when using BP combined with ZCR and FD. The ROC curve, showed excellent discrimination between responders and non-responders, with an AUC of 0.99, indicating the stability of the selected features for predicting treatment response. High classification accuracy suggests that machine learning-based EEG analysis might be promising to provide a personalized guideline for rTMS new therapy protocol.