The main issues in using hand gesture recognition models powered by electromyography (EMG) to improve motor therapy for individuals with neuromuscular injuries are covered in this work. In particular, we address the challenge of reliably classifying gestures and accurately interpreting EMG signals. We sought to determine the best strategy for this assignment by utilizing machine learning methods including Support Vector Machines (SVM), K-Nearest Neighbors (KNN), and Multilayer Perceptrons (MLP). 36 participants used the MYO Thalmic armband to perform six fundamental hand gestures, and their EMG data was recorded. Our comparative analysis revealed that the MLP model achieved the highest performance with an accuracy exceeding 94%, outperforming SVM (92%) and KNN (85%). These results demonstrate that MLP is particularly well-suited for addressing the complex classification challenges inherent in EMG signal interpretation. By incorporating these models into rehabilitation software, real-time feedback may be given, empowering patients and enhancing the effectiveness of treatment. This study not only shows that MLP performs better than other models, but it also shows how well it can deal with rehabilitation problems in the actual world. Future studies should focus on verifying these models using more modern EMG collection equipment and investigating their scalability for a variety of patient populations. Overcoming these challenges could revolutionize rehabilitation practices and give patients more accessible and reasonably priced options.