Pain determination using trend analysis, medical device incorporating machine learning, economic discriminant model, and IoT, tailormade machine learning, and novel brainwave feature quantity for pain determination
A high accuracy information extracting device construction system includes: a feature quantity extraction expression list generating unit for generating a feature quantity extraction expression list; a feature quantity calculating unit for calculating feature quantities of teacher data by means of respective feature quantity extracting expressions; a teacher data supply unit for supplying teacher data; an evaluation value calculating unit for generating information extracting expressions by means of machine learning on the basis of the calculated feature quantities of teacher data and the teacher data, and calculating evaluation values for the respective feature quantity extracting expressions; and a synthesis unit for constructing a high accuracy information extracting device using T weak information extracting parts F(X)t output from the evaluation value calculating unit 15 and confidence levels Ct corresponding thereto.
1. A computer implemented method for causing a computer to execute processing to obtain an improved differentiation model based on an accuracy of a single differentiation model to be derived by machine learning, the computer including a processor, a memory for storing the program and an interface configured to communicate with a database storing brainwave data samples of subjects respectively associated with pain labels, comprising:
a. dividing, by the processor, the brainwave data samples of the subjects into a first predetermined number of first groups;
b. determining, by the processor, a hyperparameter of a penalty term of the single differentiation model by executing k-fold cross-validation of supervised machine learning on the brainwave data samples of the first groups according to the pain labels, where k equals to the first predetermined number;
c. generating, by the processor, the single differentiation model by using the determined hyperparameter and all the brainwave data samples of the subjects to calculate first differentiation accuracy for each of the subjects;
d. ranking, by the processor, the subjects based on the first differentiation accuracy to group the subjects into a second predetermined number of second groups according to ranking order;
e. determining, by the processor, an individual hyperparameter of the penalty term of an individual differentiation model for each of the second groups by executing l-fold cross-validation of supervised machine learning on the brainwave data samples of each of the second groups, where l is a predetermined number common to the second groups;
f. generating, by the processor, the individual differentiation model for each of the second groups using the determined individual hyperparameter; and
g. identifying a differentiation maximum (MAX) model for each of the subjects among the individual differentiation models by searching for the individual differentiation model with the highest second differentiation accuracy for each of the subject.
2. The method of claim 1 , further comprising:
i) ranking, by the processor, the individual differentiation models in descending order based on the second differentiation accuracy for each of the subjects;
ii) calculating, by the processor, values of third differentiation accuracy by an ensemble method, wherein the ensemble method includes,
preparing serially ensembles of the individual differentiation models by incorporating into an initial one of the ensembles the individual differentiation model one by one according to the descending order, wherein the initial one includes the differentiation maximum (MAX) model, and
calculating each of the values of the third differentiation accuracy based on voting of the individual differentiation models included in the ensemble; and
iii) determining, by the processor, the ensemble with the highest value of the third differentiation accuracy to generate an improved differentiation model for each of the subjects.
3. The method of claim 1 , wherein the individual differentiation models include the differentiation maximum (MAX) model for each of the subjects as a main model and the rest of the individual differentiation models as a supporter model set, further comprising,
i) calculating, by the processor, values of third differentiation accuracy by an ensemble method, wherein the ensemble method includes,
preparing ensembles each including the main model and one of the individual differentiation models included in the supporter model set by combining the main model with respective ones in the supporter model set to generate the ensembles, and
calculating each of the values of the third differentiation accuracy based on voting of the individual differentiation models included in the ensemble, and
ii) selecting, by the processor, the individual differentiation model included in the ensemble with the highest value of the third differentiation accuracy, and
iii) updating the main model by setting the main model as a set of the main model and the selected individual differentiation model and updating the supporter model set by eliminating the selected individual differentiation model from the supporter model set, and
iv) repeating the steps i) to iii) until updating of the supporter model set by elimination of the selected individual differentiation model from the supporter model set terminates to determine an improved differentiation model for each of the subjects as the ensemble with the highest value of the third differentiation accuracy.
4. A non-transitory recording medium storing a program for causing a computer to execute processing to obtain an improved differentiation model based on an accuracy of a single differentiation model to be derived by machine learning, the computer including a processor, a memory for storing the program and an interface configured to communicate with a database storing brainwave data samples of subjects respectively associated with pain labels, the processing comprising:
a. dividing, by the processor, the brainwave data samples of the subjects into a first predetermined number of first groups;
b. determining, by the processor, a hyperparameter of a penalty term of the single differentiation model by executing k-fold cross-validation of supervised machine learning based on the brainwave data samples of the first groups according to the pain labels, where k equals to the first predetermined number;
c. generating, by the processor, the single differentiation model by using the determined hyperparameter and all the brainwave data samples of the subjects to calculate first differentiation accuracy for each of the subjects;
d. ranking, by the processor, the subjects based on the first differentiation accuracy to group the subjects into a second predetermined number of second groups according to ranking order;
e. determining, by the processor, an individual hyperparameter of the penalty term of an individual differentiation model for each of the second groups by executing l-fold cross-validation of supervised machine learning on the brainwave data samples of each of the second groups, where l is a predetermined number common to the second groups;
f. generating, by the processor, the individual differentiation model for each of the second groups using the determined individual hyperparameter; and
g. identifying a differentiation maximum (MAX) model for each of the subjects among the individual differentiation models by searching for the individual differentiation model with the highest second differentiation accuracy for each of the subjects.
5. A system for executing a computer-implemented method to obtain an improved differentiation model based on an accuracy of a single differentiation model to be derived by machine learning, the system comprising:
a processor;
a memory for storing a machine learning program; and
an interface configured to communicate with a database storing brainwave data samples of subjects respectively associated with pain labels;
wherein the processor configured to:
divide the brainwave data samples into a predetermined number of first groups;
determine a hyperparameter of a penalty term of the single differentiation model by executing k-fold cross-validation of supervised machine learning on the brainwave data samples of the first groups according to the pain labels, where k equals to the first predetermined number;
generate the single differentiation model by using the determined hyperparameter and all the brainwave data samples of the subjects to calculate first differentiation accuracy for each of the subjects;
rank the subjects based on the first differentiation accuracy to group the subjects into a second predetermined number of second groups according to ranking order;
determine an individual hyperparameter of the penalty term of an individual differentiation model for each of the second groups by executing l-fold cross-validation of supervised machine learning on the brainwave data samples of each of the second groups, where l is a predetermined number common to the second groups;
generate the individual differentiation model for each of the second groups using the determined individual hyperparameter; and
identify a differentiation maximum (MAX) model for each of the subjects among the individual differentiation models by searching for the individual differentiation model with the highest second differentiation accuracy for each of the subjects.