Self-Learning Neural Network for Brain Stimulation

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current brain-computer interfaces face challenges in directly recording and stimulating neural activity, limiting their ability to effectively restore or enhance brain and nervous system function, particularly in inducing neuroplasticity for rehabilitation or enhancing capabilities.

Innovation Solution

A system comprising self-learning deep recurrent neural networks that stimulate specific regions of the nervous system based on recorded neural signals, using a combination of first and second artificial networks to generate and adjust stimulation patterns, thereby inducing new neural connections and optimizing behavioral outputs.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Loss of information

If brain-computer interfaces are used to directly record and stimulate neural activity, then the ability to transfer information between brain and computer is improved, but the complexity of the system and difficulty of implementation increase significantly

Engineering Contradiction:

Improveinformation transfer capabilityVSAvoidsystem complexity

Core Design Contradiction:

Loss of informationVSDevice complexity

Solution Approach 1:

The patent introduces artificial neural networks as intermediary components that bridge the brain and computer system. These networks receive neural signals, process them through learned transformations, and generate stimulation patterns, thereby mediating the complex interaction between biological and artificial systems while reducing direct complexity requirements

Inventive Principle:

Principle #24Intermediary (Mediator)

Solution Approach 2:

The system employs self-learning artificial neural networks that automatically adapt and optimize their parameters through exposure to neural data. This self-service capability reduces the need for manual configuration and expert intervention, thereby lowering implementation difficulty while maintaining high information transfer capability

Inventive Principle:

Principle #25Self-service

2Reliability

If traditional brain-computer interfaces are used for neural stimulation, then some level of function restoration is achieved, but the ability to induce neuroplasticity and create new neural connections is limited

Engineering Contradiction:

Improvefunction restoration capabilityVSAvoidneuroplasticity induction capability

Core Design Contradiction:

ReliabilityVSAdaptability or versatility

Solution Approach 1:

The patent implements dynamic stimulation patterns generated by recurrent neural networks that adapt in real-time based on incoming neural signals. This dynamic approach allows the system to not only restore existing functions but also induce neuroplasticity by providing adaptive, context-dependent stimulation that promotes new neural connection formation

Inventive Principle:

Principle #15Dynamics

Solution Approach 2:

The artificial neural networks dynamically adjust stimulation parameters such as amplitude, frequency, and timing based on learned relationships from neural data. These parameter changes enable the system to optimize both function restoration and neuroplasticity induction, achieving versatility across different rehabilitation goals

Inventive Principle:

Principle #35Parameter changes

3Ease of manufacture

If simple stimulation patterns are used, then the system is easier to implement, but the ability to optimize behavioral outputs and induce meaningful neuroplasticity is reduced

Engineering Contradiction:

Improvesystem implementation easeVSAvoidbehavioral output optimization

Core Design Contradiction:

Ease of manufactureVSProductivity

Solution Approach 1:

The system employs self-learning neural networks that automatically optimize stimulation patterns through exposure to neural and behavioral data. This self-service learning capability eliminates the need for complex manual optimization procedures, making the system easier to implement while simultaneously achieving high-level behavioral output optimization

Inventive Principle:

Principle #25Self-service

Solution Approach 2:

The patent implements closed-loop feedback where neural signals and behavioral outputs are continuously monitored and used to train and refine the artificial neural networks. This feedback mechanism enables automatic optimization of stimulation patterns to maximize behavioral outcomes without requiring complex manual intervention

Inventive Principle:

Principle #23Feedback

Data Source

PatentUS11083895B2Systems and methods for augmenting and/or restoring brain and nervous system function and inducing new neural connections using self-learning artificial networks

Publication Date: 2021.08.10 UNIV OF WASHINGTON

AI summary

Systems and methods for restoring and/or augmenting neural function are disclosed herein. One such method includes receiving signals associated with a first region of the nervous system of the individual, and generating a stimulation pattern based on (a) the signals associated with the first region of the nervous system and (b) a first artificial network. The method can further include outputting the stimulation pattern to (a) a second region of the nervous system to induce a behavioral output from the individual and (b) a second artificial network configured to predict the behavioral output from the individual. The method can also include comparing the induced behavioral output to the predicted behavioral output to generate an error signal. Parameters of the first artificial network can be adjusted using the error signal and the second artificial network to optimize the stimulation patterns and other output signals to achieve restoration and/or augmentation goals.