Decoding Single Channel Forearm EMG Data
A project exploring how interpretable decoding models can be built from sparse forearm biosignals collected under constrained conditions.
Project Overview
This investigation asks whether meaningful information can be recovered from a single-channel forearm EMG signal using methods that are both computationally lightweight and interpretable. The motivation is to study how much structure is available in low-dimensional biosignal data and how it can be exploited for practical decoding tasks.
Methods
The analysis pipeline combines signal conditioning, feature extraction, and decoding algorithms to relate short time windows of EMG activity to underlying movement or control intent. The emphasis is on transparent methods that make it easier to understand what features are driving the model decisions.
Results
The work demonstrates that even sparse EMG signals can contain enough structure to support informative decoding when the experimental setting and modelling choices are carefully matched. The results also clarify the limits of simple models and suggest a clear path for future refinement.