Reference stimulus

1 . A computer implemented method for building a machine learning model and for differentiating and diagnosing a pain level of an object during a surgery by standardizing the pain level, the computer including a memory, processor and an interface connected to an electroencephalograph, the method comprising:

acquiring, by the electroencephalograph, a plurality of brainwave data as reaction data from the object to store the brainwave data to the memory, the acquiring the plurality of brainwave data including:

acquiring a first brainwave data from the object being in a first state; and

acquiring a second brainwave data from the object being in a second state; and

building, by the processor, the machine learning model unique to the object for differentiating the pain level of the object on the first and the second brainwaves data,

wherein the first brainwave data of the object are measured by using a stimulation application device in the first state with applying different levels of pain stimulation in a step-like manner to the object to store the first brainwave data to a storage device and the second brainwave data of the object are measured in the second state with applying no pain to the object to store the second brainwave data to the storage device and wherein the building the machine learning model includes:

I) the step of creating COVAS (computerized visual analog scale) template for standardizing the pain level, wherein the step of creating COVAS template includes:

a) the step of performing pain tests on a plurality of healthy subjects to acquire a plurality of COVAS data with applying different levels of the pain stimulation in a step-like manner to each of the healthy subjects by using the stimulation application device;

b) the step of averaging the plurality of COVAS data in each level of the pain stimulation to create COVAS templates, wherein each of the COVAS templates associates the averaged COVAS data with the corresponding level of the pain stimulation as a standardized label and includes information for showing a first trigger timing of initiation of the pain stimulation in the COVAS template; and

II) the step of building the machine learning model for predicting a pain score by differentiating the pain level of the object, wherein the step of building the machine learning model includes:

c) the step of reading out the first and the second brainwave data from the storage device;

d) the step of cutting out the first and the second brainwaves data thereof based on the COVAS templates each corresponding to the level of the pain stimulation by matching the first trigger timing with a second trigger timing of initiation of the pain stimulation in the first and the second brainwaves data; and

e) the step of learning the cut out brainwaves data thereof and the standardized label of the COVAS template corresponding to each of the cut out brainwaves data thereof as training data to create the machine learning model; and

the method further comprising predicting, by the processor, the pain score of the object as an output of the machine learning model which receives as an input brainwave data collected from the object during the surgery.

2 . The method of claim 1 , wherein the step of learning the cut out brainwaves data or analysis data thereof includes:

the step of preprocessing the cut out brainwaves data by extracting feature values of absolute amplitude, entropy, frequency power from predetermined frequency bands and coherence from the brainwaves data of each brainwave measurement channel,

the step of performing a sample augmentation method to increase the number of samples, and

the step of creating, by using the increased samples, the machine learning model of LSTM (Long Short Term Memory) to be fitted to the object.

3 . The method of claim 2 , wherein the step of learning the cut out brainwaves data or analysis data thereof further includes:

the step of sorting COVAS templates in ascending order from the minimum value of zero to the maximum value,

the step of cutting out, from the sorted COVAS templates, a predetermined number of ranges from the minimum value to the maximum value in a predetermined unit,

the step of calculating standardization parameters for searching the machine learning model from the extracted feature values based on the mean value and the standard deviation of each of the predetermined number of ranges, and

wherein the step of creating the machine learning model of LSTM includes the step of performing ensemble learning by using the standardization parameters.

4 . A recording medium on which a program for building a machine learning model and for differentiating and diagnosing a pain level of an object during a surgery by standardizing the pain level, wherein when the program is executed by a computer system equipped with a processor, the program causes the processor to execute the method of claim 1 .

5 . A system for building a machine learning model and for differentiating and diagnosing a pain level of an object during a surgery by standardizing the pain level, comprising:

an acquisition means of acquiring, by the electroencephalograph, a plurality of brainwave data as reaction data from an object, the acquisition means including a first means for acquiring a first brainwave data from the object being in a first state and a second means for acquiring a second brainwave data from the object being in a second state; and

a machine learning model building means for building a machine learning model unique to the object for differentiating a pain level of the object, based on the first and the second brainwaves data;

wherein the first brainwave data of the object are measured by using a stimulation application device in the first state with applying different levels of pain stimulation in a step-like manner to the object to store the first brainwave data to a storage device and the second brainwave data of the object are measured in the second state with applying no pain to the object to store the second brainwave data to the storage device and wherein the machine learning model building means includes:

I) a COVAS (computerized visual analog scale) template creating means for standardizing the pain level, wherein the COVAS template creating means includes:

a) means for performing pain tests on a plurality of healthy subjects to acquire a plurality of COVAS data with applying different levels of the pain stimulation in a step-like manner to each of the healthy subjects by using the stimulation application device;

b) means for averaging the plurality of COVAS data in each level of the pain stimulation to create COVAS templates, wherein each of the COVAS templates associates the averaged COVAS data with the corresponding level of the pain stimulation as a standardized label and includes information for showing a first trigger timing of initiation of the pain stimulation in the COVAS template; and

II) a machine learning model building means for building the machine learning model to predict a pain score by differentiating the pain level of the object, wherein the machine learning model building means includes:

c) means for reading out the first and the second brainwave data from the storage device;

d) means for cutting out the first and the second brainwaves data thereof based on the COVAS templates each corresponding to the level of the pain stimulation by matching the first trigger timing with a second trigger timing of initiation of the pain stimulation in the first and the second brainwaves data; and

e) means for learning the cut out brainwaves data thereof and the standardized label of the COVAS template corresponding to each of the cut out brainwaves data thereof as training data to create the machine learning model; and

the system further comprising predicting means for predicting the pain score of the object as an output of the machine learning model which receives as an input brainwave data collected from the object during the surgery.