Physiological Signal Fusion Using Probabilistic Reliability Weighting
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing subject monitoring systems are prone to inaccuracies due to motion artifacts, leading to unreliable physiological signal readings.
Innovation Solution
A computer-implemented method using probabilistic models to process physiological signals, determining the probability distribution of true values and adjusting the contribution of each measure set to an output signal based on reliability, allowing for weighted combination or selection of signals to generate a more accurate and robust output.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional subject monitoring systems are used to generate physiological signals, then the system provides basic monitoring capability, but the signals are prone to inaccuracies due to motion artifacts and other interference
Solution Approach 1:
The patent introduces probabilistic models as an intermediary layer between the raw physiological signals and the output. These models act as mediators that process the uncertain signal data through probability distributions, allowing the system to handle measurement uncertainty mathematically and produce more reliable output signals that account for motion artifacts and other interference
Solution Approach 2:
The patent transforms the physiological signal processing from deterministic value handling to probabilistic parameter representation. By defining probability distributions for signal measures and processing these probabilistic parameters through mathematical operations, the system can better represent uncertainty and improve both reliability and measurement precision simultaneously
2Loss of information
If multiple physiological signals are combined to improve monitoring accuracy, then the information completeness increases, but the complexity of processing and determining contribution weights increases
Solution Approach 1:
The patent processes multiple physiological signals and their probability distributions, allowing partial contribution from each signal based on its reliability. The system can selectively weight or exclude signals as needed, processing more information than a single signal would provide while maintaining manageable complexity through the probabilistic framework that naturally handles signal integration
Data Source
Figure 1~2
Figure 3~4
Figure 5~6
AI summary
A mechanism for generating an output signal from one or more physiological signals, where the contribution of a measure set of each physiological signal to the output signals is dependent upon a respective probabilistic model for the physiological signal.