This project presents an automated Electromyogram (EMG) Diagnosis System for Rehabilitation using Support Vector Machine (SVM) technology to improve the diagnosis of neuromuscular conditions. EMG is a valuable tool for analyzing muscle activity, but traditional methods often rely on manual interpretation, which can be subjective and slow, especially in healthcare settings with limited specialized personnel. The system aims to address these issues by providing a real-time, machine learning-based solution.

The core objective is to develop a system that collects EMG data, processes it to remove noise, and uses machine learning models like SVM and Convolutional Neural Networks (CNN) to classify muscle function as healthy or unhealthy. Initial testing revealed that SVM achieved an accuracy of 80%, while CNN reached only 40%, highlighting SVM's suitability for small datasets.

The project involved designing and simulating an EMG circuit using Proteus and collecting real data from five participants (three healthy and two with neuromuscular conditions) at the UMS Rehabilitation Center using the Delsys Trigno Lite system. The system was also tested on an online EMG dataset, where CNN achieved an accuracy of 87%.

In conclusion, the SVM-based EMG system offers a practical solution for improving diagnosis and rehabilitation, particularly in clinical settings with limited data. The project demonstrates SVM's effectiveness with smaller datasets, while CNN shows potential for larger datasets, providing a faster and more accurate diagnostic process for neuromuscular disorders.