Method for prediction of cortical spiking trains

A method for prediction of encephalic pulse neutral signal of a subject, characterized in that the method comprises the following steps in the following order:

• 1) collecting original encephalic pulse neural signal from the subject by using micro electrode array;
• dividing the original encephalic pulse neutral signal into a plurality of time slots by using 10 milliseconds as a time slot breadth;
• designating the time slot with existence of the original pulse neural signal in the time slot as 1 and designating the time slot without existence of the original pulse neural signal in the time slot as 0 to complete discretization;
• dividing each pulse neural signal channel into 50% training set, 20% verification set and 30% test set respectively along a time axis;
• checking whether pulse release rate in the training set, the verification set and the test set of the signal channel is between 2 Hz and 40 Hz;
• screening out signal channels with pulse release rate that is not between 2 Hz and 40 Hz;
• 2) using mutual information to obtain numerous anterior encephalic signal channels with the maximum value of mutual information with each signal channel in posterior encephalic region as an independent variable of a model of posterior encephalic pulse neutral signal generation probability;
• after that, setting sliding window in view of each time slot in all posterior encephalic signal channels to be predicted to select input independent variable of the model;
• once the input independent variable of the model is obtained, proceeding with modeling of posterior encephalic pulse neutral signal generation probability by using Nonhomogeneous Poisson point process generalized linear model;
• 3) using the model in Step 2) to obtain a sequence for pulse release probability of pulse neural signal to be predicted in the posterior encephalic region;
• using Discrete Time Rescaling Kolmogorov-Smirnov Statistics as a dimension for measurement of model effect to calculate model effect value;
• 4) taking the model effect measurement value obtained in Step 3) as a target function for derivation to obtain a numerical gradient;
• proceeding with optimized calculation of optimal parameters with Quasi-Newton Method;
• 5) executing Step 3) and 4) iteratively when calculating optimal parameters until accuracy of the model corresponding to the optimized parameters is up to a local limit point or upper limit of iterations;
• introducing final parameters into the model, and using the model to complete prediction of posterior encephalic pulse neutral signal;
• and
• 6) outputting predicted posterior encephalic pulse neutral signal through a display device.