Audio Mobility Analysis for Fall Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current mobility monitoring technologies for elderly individuals, particularly those with dementia, are inadequate in predicting falls and providing effective prevention, as they often rely on invasive methods or struggle with noise interference in home settings, raising privacy concerns and lacking reliability.
Innovation Solution
A method using audio signals from microphones to classify footstep regions and analyze cadence, hesitancy, and balance through supervised and unsupervised learning algorithms, determining a mobility factor to provide feedback and predict potential falls.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If video cameras are placed in rooms for non-contact monitoring, then mobility and gait measurement is improved, but privacy concerns and user reluctance increase
Solution Approach 1:
The patent replaces optical/mechanical video camera systems with an acoustic sensing system using microphones and audio signal processing. This substitution maintains the ability to measure mobility parameters (cadence, stride length, velocity) while eliminating the privacy intrusion of visual monitoring, as audio data is less personally revealing and more acceptable to users.
Solution Approach 2:
The patent introduces audio signals as an intermediary medium to indirectly measure mobility without direct visual observation. Instead of cameras capturing images of the subject, microphones capture sound waves generated by footstep events, which then serve as the basis for deriving mobility metrics through signal processing and machine learning algorithms.
2Reliability
If traditional acoustic event analysis is used in home settings, then fall detection capability is improved, but performance deteriorates due to interfering acoustic noise
Solution Approach 1:
The patent transforms the acoustic signal from raw time-domain waveforms into frequency-domain representations using Fourier transforms and spectrograms. This parameter transformation allows the system to identify footstep events based on their characteristic frequency patterns, which remain distinguishable from background noise even in noisy home environments. The system also extracts multiple acoustic features (energy, zero-crossing rate, spectral centroid) to improve detection robustness.
Solution Approach 2:
The patent performs preliminary classification of audio regions using supervised machine learning algorithms before attempting fall detection. The system pre-identifies footstep events and filters out non-footstep sounds through trained classifiers, creating a cleaned dataset of genuine footstep sequences. This preliminary action removes interfering noise and false positives before the actual fall detection analysis, significantly improving reliability in noisy environments.
3Ease of operation
If basic infrared motion sensors are used for activity detection, then non-contact monitoring is achieved, but information about mobility changes and fall likelihood is insufficient
Solution Approach 1:
The patent transitions from the spatial dimension detection of infrared motion sensors to the temporal and frequency dimension analysis of acoustic signals. While infrared sensors detect movement in space, the acoustic system analyzes sound waves over time and frequency, extracting rich temporal patterns (cadence, rhythm, hesitation) and spectral characteristics that provide detailed information about mobility status and fall risk that simple motion detection cannot capture.
Solution Approach 2:
The patent extracts multiple parameters from acoustic signals including cadence, stride length, walking velocity, hesitancy, and balance metrics. These parameters are derived through sophisticated signal processing and machine learning analysis of footstep sequences, providing comprehensive mobility assessment that goes far beyond the simple presence/absence detection of infrared sensors.
4Reliability
If existing fall detection technologies are deployed, then fall detection efficiency is improved, but ethical issues regarding liberty and privacy worsen
Solution Approach 1:
The patent replaces invasive monitoring systems (such as wearables, cameras, or direct contact sensors) with non-invasive acoustic sensing. This substitution maintains fall detection efficiency by analyzing footstep patterns and mobility changes, while simultaneously reducing ethical concerns as the system does not require physical contact with or direct visual monitoring of the subject, thereby better preserving liberty and privacy.
Data Source
AI summary
A method for measuring the mobility of a subject, the method comprising: receiving an audio signal from one or more microphones: for each of a plurality of overlapping regions of the audio signal, classifying the region as containing the sound of a footstep using a first supervised learning algorithm, determining that two or more of the regions classified as containing the sound of a footstep correspond to a series of two or more consecutive footsteps of a subject; and using a first neural network, analysing the determined two or more regions to determine a mobility factor.