Neural interface

Abstract

Method(s) and apparatus are provided for interfacing with a nervous system of a subject. In response to receiving a plurality of neurological signals associated with the neural activity of the first portion of nervous system: processing neural sample data representative of the received plurality of neurological signals using a first one or more machine learning (ML) technique(s) trained for generating estimates of neural data representative of the neural activity of the first portion of nervous system; and transmitting data representative of the neural data estimates to a first device associated with the first portion of nervous system; and in response to receiving device data from a second device associated with a second portion of the nervous system: generating one or more neurological stimulus signal(s) by inputting the received device data to a second one or more ML technique(s) trained for estimating one or more neurological stimulus signal(s) associated with the device data for input to the second portion of nervous system; and transmitting the one or more estimated neurological stimulus signal(s) towards the second portion of nervous system of the subject.

1 . A computer implemented method for interfacing with a nervous system of a subject, the method comprising:
 in response to receiving a plurality of neurological signals associated with the neural activity of the first portion of nervous system, performing the steps of:   processing neural sample data representative of the received plurality of neurological signals using a first one or more machine learning (ML) technique(s) trained for generating estimates of neural data representative of the neural activity of the first portion of nervous system; and   transmitting data representative of the neural data estimates to a first device associated with the first portion of nervous system; and   in response to receiving device data from a second device associated with a second portion of the nervous system, performing the steps of:   generating one or more neurological stimulus signal(s) by inputting the received device data to a second one or more ML technique(s) trained for estimating one or more neurological stimulus signal(s) associated with the device data for input to the second portion of nervous system; and   transmitting the one or more estimated neurological stimulus signal(s) towards the second portion of nervous system of the subject.   
     
     
         2 .- 160 . (canceled) 
     
     
         161 . The computer implemented method as claimed in  claim 1 , wherein the estimates of neural data representative of neural activity as generated or calculated by at least one of the ML techniques are associated with one or more bodily variables. 
     
     
         162 . The computer implemented method of  claim 1 , further comprising:
 receiving at least one set of performance data associated with the first one or more ML technique(s) or the second one or more ML technique(s);   evaluating the set of performance data to determine whether to retrain the first one or more ML technique(s) or the second one or more ML technique(s); and   retraining the first one or more ML technique(s) in response to determining to retrain the first one or more ML technique(s) or the second one or more ML.   
     
     
         163 . The computer implemented method of  claim 1 , wherein the first portion of the nervous system comprises a first plurality of neurons of the subject clustered around multiple neural receivers, each neural receiver configured for outputting neurological signals associated with neural activity on one or more of the plurality of neurons, the method comprising:
 receiving one or more neurological signals from the neural receivers associated with the plurality of neurons of the subject; and   classifying the one or more neurological signals into one or more categories of neural data using at least one of the first one or more ML technique(s).   
     
     
         164 . The computer implemented method of  claim 1 , further comprising:
 generating neural sample data representative of the neurological signals by capturing samples of the neurological signals when neural activity is detected; and   processing the neural sample data using at least one of the first one or more ML technique(s) to generate neural data representative of neural information associated with the neural activity.   
     
     
         165 . The computer implemented method as claimed in  claim 164 , further comprising generating a training set of neural sample data by:
 storing captured neural sample data received from the plurality of neurological signals, wherein the neural sample data is timestamped;   capturing and storing sensor data from one or more sensors trained on the subject, wherein the sensor data is timestamped;   synchronising the neural sample data with the sensor data; and   identifying portions of the neural sample data associated with neural activity;   determining neural data labels for each identified portion of neural sample data by analysing portions of the sensor data corresponding to the identified portion of neural sample data;   labelling the identified portions of neural sample data based on the determined neural data labels; and   storing the labelled identified portions of neural sample data as the training set of neural sample data.   
     
     
         166 . The computer implemented method of  claim 1  further comprising training at least one of the first one or more ML technique(s) based on a training set of neural sample data, wherein each neural sample data in the training set is labelled associated with a neural data label identifying the neural data contained therein. 
     
     
         167 . The computer implemented method of  claim 1 , wherein at least one of the first one or more ML technique(s) comprise at least one or more ML technique(s) or combinations thereof from the group of:
 a) neural networks;   b) Hidden Markov Models;   c) Gaussian process dynamics models;   d) autoencoder/decoder networks;   e) adversarial/discriminator networks;   f) convolutional neural networks;   g) long short term memory neural networks; and   h) any other ML or classifier/classification technique or combinations thereof suitable for operating on said received neurological signal(s).   
     
     
         168 . The computer implemented method of  claim 1 , wherein at least one of the first one or more ML technique(s) is based on a neural network autoencoder structure, the neural network autoencoder structure comprising an encoding network and a decoding network, the encoding network comprising one or more hidden layer(s) and the decoding network comprising one or more hidden layer(s), wherein the neural network autoencoder is trained to output a neural data label vector that is capable of classifying each portion of neural sample data from a training set of neural sample data into one or more neural data labels, the method comprising:
 inputting neural sample data to the autoencoder for real-time classification of neurological signals.   
     
     
         169 . The computer implemented method as claimed in  claim 168 , the method further comprising:
 training the neural network autoencoder for outputting a neural data label vector that is capable of classifying each portion of neural sample data from a training set of neural sample data into one or more neural data labels; and   using the trained weights of the hidden layer(s) of the autoencoder for real-time classification of neurological signals.   
     
     
         170 . A computer implemented method for determining neural activity of a portion of a nervous system of a subject, the method comprising:
 receiving a plurality of neurological signals associated with the neural activity of the portion of the nervous system; and   processing neural sample data representative of the received plurality of neurological signals using one or more machine learning (ML) technique(s) trained for generating estimates of neural activity or combinations thereof associated with the neural activity of the portion of nervous system; and   transmitting data representative of the neural activity estimates to a device for performing operations based on the neural activity estimate(s).   
     
     
         171 . The computer implemented method as claimed in  claim 170 , wherein the neural activity comprises neural activity encoding one or more bodily variable(s) of the portion of the nervous system of the subject, the method further comprising:
 processing neural sample data representative of the received plurality of neurological signals using one or more machine learning (ML) technique(s) trained for generating estimates of one or more bodily variables or combinations thereof associated with the neural activity of the portion of nervous system; and   transmitting data representative of the one or more bodily variable estimates to a device for performing operations based on the bodily variable estimate(s).   
     
     
         172 . The computer implemented method of  claim 170 , wherein the portion of the nervous system comprises a plurality of neurons of the subject clustered around multiple neural receivers, each neural receiver configured for outputting neurological signals associated with neural activity on one or more of the plurality of neurons, the method comprising:
 receiving one or more neurological signals from the neural receivers associated with the plurality of neurons of the subject; and   classifying the one or more neurological signals into one or more categories of bodily variable(s) using the one or more ML technique(s).   
     
     
         173 . The computer implemented method of  claim 170 , further comprising:
 generating neural sample data representative of the neurological signals by capturing samples of the neurological signals when neural activity encoding one or more bodily variable(s) is detected; and   processing the neural sample data using the one or more ML technique(s) to generate data representative of bodily variable estimates.   
     
     
         174 . The computer implemented method of  claim 170 , further comprising generating a training set of neural sample data by:
 storing captured neural sample data received from the plurality of neurological signals, wherein the neural sample data is timestamped;   capturing and storing sensor data from one or more sensors trained on the subject, wherein the sensor data is timestamped;   synchronising the neural sample data with the sensor data; and   identifying portions of the neural sample data associated with neural activity encoding one or more bodily variable(s);   determining bodily variable labels for each identified portion of neural sample data by analysing portions of the sensor data corresponding to the identified portion of neural sample data;   labelling the identified portions of neural sample data based on the determined bodily variable labels; and   generating a labelled training set of neural sample data associated with the bodily variable of interest based on the labelled identified portions of neural sample data.   
     
     
         175 . The computer implemented method as claimed in  claim 174 , wherein generating the labelled training set of neural sample data further comprises storing the labelled identified portions of neural sample data as the labelled training set of neural sample data. 
     
     
         176 . The computer implemented method of  claim 173 , further comprising analysing the detected portions of neural sample data using one or more ML technique(s) to generate a set of classification vectors associated with one or more bodily variable(s) or combinations thereof contained within detected portions of neural sample data; and
 labelling the classification vectors with bodily variable labels determined from corresponding portions of the neural sample data and sensor data.   
     
     
         178 . The computer implemented method of  claim 170  further comprising training one or more ML technique(s) based on a training set of neural sample data, wherein each neural sample data in the training set is labelled associated with a bodily variable label identifying the one or more bodily variables contained therein. 
     
     
         179 . The computer implemented method of  claim 170 , wherein the one or more ML technique(s) comprise at least one or more ML technique(s) from the group of:
 a) neural networks;   b) Hidden Markov Models;   c) Gaussian process dynamics models;   d) autoencoder/decoder networks;   e) adversarial/discriminator networks;   f) convolutional neural networks;   g) long short term memory neural networks; and   h) any other ML or classifier/classification technique or combinations thereof suitable for operating on said received neurological signal(s).