Predicting Finger Flexions from ECoG Signals

A neural-decoding project predicting continuous finger flexion directly from electrocorticographic (ECoG) recordings.

• Used a multi-layer perceptron regressor that accounts for the finger's previous motion to predict flexion from intracranial recordings.
• Benchmarked against standard ML models using Mean Squared Error and correlation between true and predicted flexions.
• Achieved a final correlation score of r = 0.4659.