Energy-Efficient Adaptive Neural Stimulator With Waveform Prediction by Sub-Threshold Interrogation of the...

Area of Science:

• Biomedical Engineering
• Neuroscience
• Implantable Devices

Background:

• Conventional neural stimulators face challenges with power consumption and efficiency.
• Existing systems often lack intelligent, adaptive stimulation capabilities.
• Implantable devices require ultra-low power operation for long-term use.

Purpose of the Study:

• To develop an implantable low-power neural stimulator with on-chip edge-learning capabilities.
• To enhance energy efficiency by minimizing power loss during stimulation.
• To enable intelligent, safe, and remotely controllable neural modulation.

Main Methods:

• Utilized subject-specific edge-learning of electrode-tissue voltage profiles to predict stimulation waveforms.
• Implemented a custom switched-capacitor output stage to reduce power loss.
• Integrated an ultra-low-power microcontroller for on-chip learning and prediction.
• Powered the system via wireless inductive energy transfer and remote control through WiFi.

Main Results:

• Achieved up to 20% reduction in power loss compared to dynamic power supply scaling.
• Demonstrated up to 3.63× lower power consumption than conventional constant-current output stages.
• Successfully validated through in vivo rat peripheral nerve stimulation, in vitro saline tests, and benchtop experiments.

Conclusions:

• The proposed intelligent neural interface system offers significant advancements in energy efficiency and safety.
• The system's remote controllability and on-chip intelligence pave the way for next-generation neural implants.
• This technology holds potential for sophisticated neural organ modulation and therapeutic applications.