Eye-Gaze and Physiological Signal Correlation for User-State Assessment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing technologies struggle to effectively associate physiological signals with user behavior and assess user states, requiring expert knowledge and lacking sufficient correlation with positions of a line of sight.
Innovation Solution
A method involving a machine-learning algorithm that processes positions of a line of sight and physiological signals, using dynamically determined latencies and durations to generate predicted values reflective of psychological constructs like emotions, cognitive load, stress, attention, and flow, by correlating subsets of data from these signals.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If physiological signals are processed using traditional methods, then expert knowledge is required to extract meaningful information, but this increases the complexity and difficulty of the analysis process
Solution Approach 1:
The patent introduces an intermediary computational model that acts as a mediator between raw physiological signals and user state assessment. This model translates complex physiological data into meaningful metrics without requiring expert knowledge to perform the analysis directly, thereby reducing processing complexity while maintaining assessment accuracy.
Solution Approach 2:
The patent replaces manual expert analysis (mechanical system) with automated machine learning algorithms. The system automatically processes physiological signals, determines latency and duration parameters, and generates predicted user states without human intervention, substituting the expert knowledge requirement with computational automation.
2Reliability
If physiological signals are analyzed without considering latency and duration, then processing is simpler, but the correlation with user behavior and line of sight positions is insufficient
Solution Approach 1:
The patent applies dynamics by making the latency and duration parameters adaptive rather than fixed. The system dynamically determines these parameters based on the specific physiological signal being analyzed and the target user state, allowing the processing window to adjust automatically to optimize correlation while managing complexity through algorithmic adaptation.
Solution Approach 2:
The patent changes key parameters (latency and duration) to optimize the correlation between physiological signals and user behavior. By varying these temporal parameters based on signal characteristics and target states, the system improves measurement reliability without requiring complex manual tuning, as the parameters are determined algorithmically.
3Measurement precision
If multiple physiological signals are processed with dynamic latency and duration, then prediction accuracy improves, but processing time and computational resources increase
Solution Approach 1:
The patent performs preliminary action by pre-determining optimal latency and duration parameters based on the type of physiological signal and target user state before actual data processing. This preparation step allows the system to use fixed, optimized parameters during runtime, reducing computational overhead and processing time while maintaining high prediction accuracy.
Solution Approach 2:
The patent segments the processing task by handling different physiological signals independently with signal-specific latency and duration parameters. This segmentation allows the system to optimize processing for each signal type separately, reducing overall computational complexity and processing time compared to a unified approach, while maintaining accuracy through specialized parameter sets.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A system for and a method of processing signals sensed from a user. The method comprises accessing positions of a line of sight of the user over a time frame, a first set of data associated with a first physiological signal and a second physiological signal. The method further comprises executing, by a processor, for at least one position of the positions of the line of sight of the user, identifying a first subset of data from the first set of data, identifying a second subset of data from the second set of data, associating the at least one position with the first subset of data and the second subset of data and causing to generate, by a machine- learning algorithm, a predicted value reflective of a pattern associated with the user. The method also comprises storing the predicted value associated with the at least one position.