Towards closed-loop deep brain stimulation: decision tree-based essential tremor patient's state classifier and...

Area of Science:

• Neurology
• Biomedical Engineering
• Signal Processing

Background:

• Progressive neurological movement disorders like Essential Tremor (ET) are often treated with Deep Brain Stimulation (DBS).
• Current DBS systems are open-loop, with fixed parameters, limiting adaptive treatment.
• Advanced ET stages require effective and responsive therapeutic interventions.

Purpose of the Study:

• To develop and evaluate a Decision Tree (DT) based algorithm for adaptive Deep Brain Stimulation (DBS) in Essential Tremor (ET) patients.
• To predict the reappearance of tremor during DBS-OFF periods using non-invasive physiological signals.
• To enhance the responsiveness of DBS therapy by optimizing ON/OFF switching based on tremor prediction.

Main Methods:

• Utilized surface electromyography (EMG) and accelerometer signals during DBS-OFF periods.
• Developed a Decision Tree (DT) algorithm to classify patient state and predict imminent tremor.
• Implemented a closed-loop control strategy where DBS is reactivated upon predicted tremor onset.

Main Results:

• The algorithm achieved an overall accuracy of 93.3% in classifying patient state and predicting tremor.
• Demonstrated high sensitivity (97.4%) with a low false alarm rate (2.9%).
• The predicted tremor reappearance closely matched actual onset, with a prediction-to-actual delay ratio of approximately 0.93.

Conclusions:

• The developed Decision Tree algorithm effectively predicts tremor recurrence in Essential Tremor patients.
• This predictive capability allows for more responsive and potentially optimized Deep Brain Stimulation therapy.
• The non-invasive signal-based approach offers a promising method for adaptive control of DBS systems.