Autonomous optimization of neuroprosthetic stimulation parameters that drive the motor cortex and spinal cord outputs in rats and monkeys

Cell Reports Medicine, 2023 · DOI: https://doi.org/10.1016/j.xcrm.2023.101008 · Published: April 18, 2023

Simple Explanation

This paper introduces a self-driving algorithm for intelligent neuroprostheses. The algorithm autonomously explores and selects the best parameters of stimulation delivered to the nervous system to evoke movements in real time in living subjects. The algorithm can rapidly solve high-dimensionality problems faced in clinical settings, increasing neuromodulation treatment time and efficacy. The core of the approach is Gaussian-Process (GP)-based Bayesian Optimization (BO). GP-BO algorithms iteratively test single-input parameter combinations leveraging responses to build an evidence-based approximation that describes how stimulus choices affect the desired output. The implemented GP-BO-based algorithm identifies optimal stimulation parameters for desired motor output, defined as a scalar value function subject to optimization. Responses are captured with sensing techniques like electromyographic (EMG) activity and movement kinematics with cameras.

Study Duration

Multiple weeks

Participants

Rats, capuchin monkeys, and macaque monkeys

Evidence Level

Level 2: Experimental study with animal models

Key Findings

1
GP-BO efficiently explores the neurostimulation space, outperforming other search strategies after testing only a fraction of the possible combinations.
2
GP-BO optimization can be applied across diverse biological targets (brain, spinal cord), animal models (rats, non-human primates), in healthy subjects, and in neuroprosthetic intervention after injury.
3
GP-BO can embed and improve ‘‘prior’’ expert/clinical knowledge to dramatically enhance its performance.

Research Summary

The study leverages an algorithmic class, Gaussian-process (GP)-based Bayesian optimization (BO), to solve the problem of handling optimization in large parameter spaces in neural stimulation. The results demonstrate optimization across diverse biological targets (brain, spinal cord), animal models (rats, non-human primates), in healthy subjects, and in neuroprosthetic intervention after injury, for both immediate and continual learning over multiple sessions. The study advocates for broader establishment of learning agents as structural elements of neuroprosthetic design, enabling personalization and maximization of therapeutic effectiveness.

Practical Implications

Personalized Neuroprosthetic Design

GP-BO allows for user-specific personalization of neurostimulation in real time, relieving the experimental and clinical burden of implementation.

Improved Treatment Efficacy

The algorithm optimizes stimulation within a limited number of queries, purely online, starting with no previous knowledge, and maintains continual learning.

Advanced Neurorehabilitation

GP-BO contributes to the development of a next generation of intelligent neural prostheses, autonomously improving movement after paralysis.

Study Limitations

1
GP-BO and autonomous learning agents will have to prove direct clinical and research adoption.
2
In all experiments we targeted single-output learning and we have not assessed how designing objective functions for the optimization of stimulation can play a major role.
3
The number of alternative ‘‘benchmark’’ algorithms to perform online neurostimulation optimization tested in our study was limited.

Your Feedback

Was this summary helpful?

Back to Neurology