Fall Detection Using Machine Learning on Wireless Sensors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional wireless sensor devices struggle to accurately distinguish between falls and activities of daily living due to the use of simple acceleration thresholds, leading to false positive detections and inaccuracies when attached in unknown orientations.
Innovation Solution
A wireless sensor device employing machine learning, specifically using a support vector machine (SVM) algorithm trained with tri-axial acceleration data, to classify falls and non-falls, with calibration vectors obtained through manual or implicit methods, allowing for accurate fall detection regardless of device orientation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If simple acceleration thresholds are used for fall detection, then the device complexity is reduced, but the measurement precision and reliability deteriorate due to inability to discriminate falls from activities of daily living
Solution Approach 1:
The patent transforms the detection approach by changing from simple threshold-based parameter comparison to machine learning-based parameter analysis. The system extracts multiple features from acceleration signals and uses trained classifiers to distinguish falls from activities of daily living, significantly improving detection accuracy while maintaining reasonable device complexity through efficient feature selection and lightweight machine learning models.
Solution Approach 2:
The patent replaces the mechanical threshold-comparison system with an intelligent machine learning system. Instead of using fixed acceleration thresholds that cannot adapt to different activities, the system employs trained classifiers that learn to distinguish fall patterns from normal activities, achieving high precision without proportionally increasing hardware complexity.
2Ease of operation
If conventional threshold-based detection is used, then the device operates simply, but false positive detections increase due to inability to differentiate falls from activities of daily living
Solution Approach 1:
The patent applies preliminary action by training machine learning classifiers beforehand with labeled fall and non-fall data. This pre-training allows the system to learn discriminative patterns before actual deployment, enabling it to reliably distinguish falls from activities of daily living during operation without requiring complex real-time decision logic, thus reducing false positives while maintaining operational simplicity.
3Measurement precision
If the sensor requires specific attachment orientations for accurate detection, then the measurement precision improves, but the ease of operation deteriorates due to user burden of correct placement
Solution Approach 1:
The patent achieves universality by developing a machine learning detection system that functions accurately across multiple device orientations. The trained classifiers learn orientation-invariant fall patterns, allowing the sensor to provide reliable fall detection regardless of how it is attached to the user's body. This eliminates the need for users to worry about specific placement orientations while maintaining high detection accuracy.
Data Source
AI summary
A method and system for fall detection using machine learning are disclosed. The method comprises detecting at least one signal by a wireless sensor device and calculating a plurality of features from the at least one detected signal. The method includes training a machine learning unit of the wireless sensor device using the features to create a fall classification and a non-fall classification for the fall detection. The system includes a sensor to detect at least one signal, a processor coupled to the sensor, and a memory device coupled to the processor, wherein the memory device includes an application that, when executed by the processor, causes the processor to calculate a plurality of features from the at least one detected signal and to train a machine learning unit of the wireless sensor device using the features to create a fall classification and a non-fall classification for the fall detection.