Brain Activity Feature Extraction for Health Discrimination
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing brain activity feature amount extraction methods are limited in their ability to accurately discriminate between healthy and non-healthy individuals due to the omission of useful determination criteria, leading to inadequate discrimination between the two groups.
Innovation Solution
A method that involves assigning tasks to subjects, measuring cerebral blood flow rate changes using near-infrared spectroscopy, calculating index values for task, measurement region, and feature amounts, and displaying the most effective data items for discrimination, while creating models with varying numbers of data items to determine the required number for accurate classification.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If the sequential selection method is used to select feature amounts, then the model building process is simplified, but the number of useful determination criteria is limited and discrimination accuracy deteriorates
Solution Approach 1:
The patent segments the feature amount selection process into multiple independent evaluation dimensions (discrimination power, correlation with cognitive function, statistical significance) rather than using a single sequential selection criterion. This allows comprehensive evaluation of multiple feature amounts simultaneously, preventing omission of useful determination criteria while maintaining systematic model building.
Solution Approach 2:
The patent changes the selection parameters from a single sequential criterion to multiple evaluation parameters including discrimination power, correlation coefficients, and statistical significance levels. By evaluating feature amounts across multiple parameter dimensions, the method identifies a more comprehensive set of useful determination criteria for accurate discrimination.
2Device complexity
If only a limited number of feature amounts are selected, then the data processing complexity is reduced, but useful determination criteria are omitted leading to inadequate discrimination
Solution Approach 1:
The patent performs preliminary evaluation of all candidate feature amounts across multiple dimensions (discrimination power, correlation, statistical significance) before final model construction. This preliminary comprehensive action ensures that no useful determination criteria are omitted, while the subsequent model building step manages complexity by selecting from the pre-evaluated feature amounts.
Solution Approach 2:
The patent replaces the mechanical sequential selection process with a multi-dimensional evaluation system that assesses feature amounts based on discrimination power, correlation coefficients, and statistical significance. This substitution allows comprehensive evaluation without proportionally increasing processing complexity, as the evaluation framework systematically handles multiple criteria simultaneously.
3Measurement precision
If multiple feature amounts are evaluated comprehensively, then discrimination accuracy is improved, but the calculation and processing time increases
Solution Approach 1:
The patent segments the comprehensive evaluation into distinct computational stages: calculating discrimination power for each feature amount, computing correlation coefficients with cognitive function, and performing statistical significance tests. This segmentation allows efficient processing of each dimension separately while achieving comprehensive evaluation, reducing overall processing time compared to unsegmented analysis.
Solution Approach 2:
The patent implements partial evaluation by focusing on the most critical dimensions (discrimination power, correlation, statistical significance) rather than exhaustively analyzing all possible feature characteristics. This partial action approach achieves sufficient discrimination accuracy without the excessive processing time that would result from comprehensive analysis of all possible features.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach effectively increases the number of determination criteria for discriminating between healthy and non-healthy individuals, suppressing the occurrence of feature amount omission and improving the accuracy of discrimination by using a combination of data items with high effectiveness and correct answer rates.
Implementation Method 1
measuring a change in a cerebral blood flow rate for each measurement region of the subject when the assignment is given to the subject
Data Source
AI summary
This brain activity feature amount extraction method includes acquiring a combination of a plurality of data items in order of higher effectiveness for discriminating whether a subject is a healthy person or a non-healthy person, among the plurality of data items, on the basis of the index value serving as the index indicating effectiveness or non-effectiveness for discriminating the group of healthy persons and the group of non-healthy persons from each other, and displaying the acquired combination of the plurality of data items in order of higher effectiveness.