Adaptive Movement Evaluation Using Dynamic Signal Quality Feature Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing movement evaluation technologies face reliability issues due to sensor artefacts and quality issues, such as drift in acceleration signals and environmental noise affecting air pressure measurements, which can lead to inaccurate fall detection and mobility assessment.
Innovation Solution
A method that dynamically adjusts feature sets based on signal quality, excluding features sensitive to artefacts or noise, ensuring reliable movement evaluation by using alternative sensors when primary sensor signals are unreliable.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If sensor signals are used for movement evaluation, then measurement capability is provided, but signal quality issues and artefacts reduce reliability
Solution Approach 1:
The system dynamically adjusts the feature set based on signal quality assessment. When signal quality is high, more features are used for evaluation; when signal quality degrades, the system selectively removes sensitive features while retaining robust ones, maintaining reliable movement evaluation adaptively responding to changing signal conditions
Solution Approach 2:
The system changes the parameters of the evaluation by selecting different subsets of features based on signal quality. The feature set is modified by removing features sensitive to artefacts (e.g., peak power, peak acceleration) when signal quality is poor, while retaining features that are robust to such artefacts, thereby maintaining evaluation reliability under varying signal conditions
2Loss of information
If multiple features are used for comprehensive movement evaluation, then evaluation completeness is improved, but sensitivity to artefacts and noise increases
Solution Approach 1:
The feature set is segmented into different categories based on their sensitivity to artefacts. The system identifies and separates features that are sensitive to signal quality issues (such as peak power, peak acceleration, and features derived from air pressure sensors) from those that are robust. This segmentation allows selective use of features based on current signal quality conditions
Solution Approach 2:
The composition of the feature set is made dynamic rather than static. The system adjusts which features are included in the evaluation based on real-time signal quality assessment. When signal quality is high, a comprehensive feature set is used; when signal quality degrades, sensitive features are removed while retaining robust features, maintaining both completeness and reliability
3Reliability
If dedicated PERS devices with controlled sensors are used, then sensor quality is improved, but device versatility is reduced
Solution Approach 1:
The system is designed to work universally with multiple types of electronic devices including smartphones, smartwatches, and dedicated PERS devices. By implementing signal quality assessment and adaptive feature selection, the system can accommodate variations in sensor quality across different device types without requiring device-specific configurations or limitations
Solution Approach 2:
The system adapts its operation based on the actual signal quality parameters observed from the sensor, rather than relying on predetermined device characteristics. This allows the same software application to function reliably across different device types by adjusting its feature selection based on measured signal quality rather than device identity
Data Source
Figure 1
Figure 2
Figure 3
AI summary
According to an aspect, there is provided a computer-implemented method for evaluating movement of a subject. The method comprises obtaining a first signal representing measurements of the subject from a first sensor; processing the first signal to determine a quality measure for the first signal; determining if the determined quality measure meets a first criterion; if the determined quality measure meets the first criterion, determining values for a plurality of features in a first feature set, the first feature set comprising one or more first features to be determined from the first signal, and evaluating the movement of the subject based on the values for the plurality of features in the first feature set; and, if the determined quality measure does not meet the first criterion, determining values for one or more features in a second feature set, wherein the one or more features in the second feature set are a subset of the plurality of features in the first feature set and the second feature set does not include at least one of the one or more first features in the first feature set, and evaluating the movement of the subject based on the values for the one or more features in the second feature set. A corresponding apparatus and computer program product are also provided.