EEG Cognitive Status Scoring for Early MCI Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for detecting and predicting mild cognitive impairment (MCI) are inadequate in sensitivity and specificity, particularly for under-served populations like older African Americans, and lack effective tools for early detection and prediction of cognitive decline before Alzheimer's disease onset.
Innovation Solution
A multiscale analysis of resting-state EEG data is used to develop a soft discrimination model that combines machine learning algorithms and majority voting, generating a cognitive status score for early detection and prediction of MCI, utilizing a series of classifiers and time-varying brain functional connectivity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If existing MCI detection methods are used, then detection capability is provided, but sensitivity and specificity are insufficient
Solution Approach 1:
The patent segments the detection process into multiple independent classifiers, each analyzing different features or aspects of the data. These classifiers work independently and their results are combined through ensemble methods, allowing the system to capture diverse patterns while maintaining high sensitivity and specificity for MCI detection across different populations.
Solution Approach 2:
The patent combines multiple classifiers into an ensemble system that integrates their predictions. By merging the results of several specialized classifiers through voting or weighted aggregation, the system achieves superior detection performance compared to any single classifier, improving both sensitivity and specificity for MCI detection.
2Loss of time
If early detection tools are developed, then cognitive decline prediction is enabled, but detection complexity increases
Solution Approach 1:
The patent performs preliminary analysis by training multiple specialized classifiers on different feature sets and data aspects before the actual detection. This preprocessing and preparation of classification models enables rapid early detection when needed, as the heavy computational work of model training is completed in advance.
Solution Approach 2:
The patent changes parameters by using multiple classifiers with different configurations, feature selections, and analysis methods. Each classifier is optimized for specific parameters or aspects of the data, allowing the ensemble system to achieve high detection accuracy while managing complexity through specialized modular components.
Data Source
AI summary
A method and system for assessing brain impairment includes generating electroencephalogram (EEG) signals from a plurality of electrodes, removing artifacts from the EEG signals to generate artifact reduced signals, generating current source density signals from the artifact reduced signals, performing multiscale analysis of dynamic functional connectivity of a brain based on the current source density signals using a plurality of different sized time windows, generating hard classifiers for each of the plurality of different size time windows, selecting classifiers from the hard classifiers to form selected classifiers, performing majority voting on a discrimination of normal cognition and mild cognitive impairment using the selected classifiers, generating an EEG-based health score of overall brain activity based on majority voting of the selected classifiers and generating a display corresponding to the health score of brain activity.