Biosignal Contextual Labeling for Mental State Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods fail to reliably quantify and characterize mental states, such as attention, from biosignals like EEG and fNIRS due to non-linearity, variability, and noise, making it difficult to generate instantaneous and numeric attention metrics.
Innovation Solution
A machine learning model is trained using biosignals and contextual category labels to predict comparative results, allowing for the identification of device operations based on categorical outputs, which can indicate mental states like workload or satisfaction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional signal processing techniques are used to analyze biosignals, then the processing method is simple and straightforward, but the ability to reliably quantify and characterize mental states is insufficient
Solution Approach 1:
The patent introduces contextual category labels as an intermediary element between raw biosignals and mental state quantification. These labels serve as a bridge that translates complex biosignal patterns into interpretable contextual categories, enabling reliable mental state characterization without requiring direct complex mathematical processing of the raw signals themselves
Solution Approach 2:
The patent replaces traditional mechanical signal processing techniques with machine learning models. Instead of using conventional filtering, transformation, and analysis methods, the system employs trained machine learning algorithms to automatically identify patterns and extract meaningful information from biosignals, significantly improving measurement precision for mental state quantification
2Measurement precision
If machine learning models are trained using complex training processes, then the prediction accuracy for mental states improves, but the training time and computational resources increase
Solution Approach 1:
The patent applies preliminary action by pre-training machine learning models using comprehensive training datasets before deployment. The models are trained in advance on labeled biosignal data to learn the relationships between biosignals and contextual categories, so that when deployed, they can quickly make accurate predictions without requiring extensive real-time processing or retraining
Solution Approach 2:
The patent uses partial action by training models on representative subsets of data and using transfer learning approaches. Instead of requiring exhaustive training on all possible scenarios, the system trains on carefully selected training datasets that capture the essential patterns, achieving good prediction accuracy with reduced training time and computational resources
3Reliability
If biosignals are analyzed without contextual labeling, then the analysis process is faster, but the characterization of mental states becomes unreliable
Solution Approach 1:
The patent applies segmentation by dividing the analysis process into distinct stages: biosignal acquisition, contextual category label assignment, and mental state prediction. This segmentation allows the system to systematically process biosignals with contextual information at appropriate stages, improving reliability without overwhelming complexity in any single component
Solution Approach 2:
The patent creates a universal framework where contextual category labels serve multiple functions: they provide training targets for machine learning models, serve as intermediate representations for analysis, and enable consistent characterization across different biosignal types and mental states. This multi-functionality increases reliability while managing complexity through a unified approach
Data Source
AI summary
Techniques are disclosed for defining a training data set to include biosignals and categorical labels representative of a context. For example, a categorical label may indicate whether a user was performing a difficult or easy mental task while the biosignal was being recorded. A set of first layers in a neural network can be trained using a portion of the training data set associated with a first set of users and at least one second layer can be trained using a portion of the training data set associated with a particular other user. The neural network can then be used to process other biosignals from the particular other user to generate predicted categorical context labels.