Interferometric Micro-Doppler Radar for 3D Gait Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current micro-Doppler radar systems are inadequate for effectively monitoring gait in individuals with Alzheimer's disease or at risk of Alzheimer's, as they fail to capture realistic human gait movements in three-dimensional space and lack robust algorithms to estimate gait patterns, especially those that are indescribable.
Innovation Solution
The development of an interferometric micro-Doppler radar system combined with deep learning artificial intelligence, which captures signals from both radial and transversal movements in three-dimensional space, using cross-talk deep models to integrate classification results from feature-based classifiers and deep learning AI for accurate gait analysis and Alzheimer's disease risk quantification.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional micro-Doppler radar is used for gait estimation, then radial movement can be detected, but transversal movement in 3D space cannot be captured
Solution Approach 1:
The patent applies interferometric technology to extend conventional 1D radial MDR into 3D space, enabling detection of both radial and transversal movements. The interferometric radar uses multiple antennas to create baseline separation, allowing measurement of angular velocity and transversal velocity components that were previously undetectable, thus resolving the dimensionality limitation
2Ease of operation
If non-wearable sensor systems are used for gait monitoring, then patient compliance is improved, but the system requires controlled research facilities and expensive floor mats
Solution Approach 1:
The interferometric MDR system is designed to be a standalone ambient sensor that automatically monitors gait without requiring patient activation or compliance actions. The system passively captures radar signals from moving bodies and processes them through interferometric correlation algorithms, eliminating the need for patient cooperation while removing the requirement for controlled facilities and expensive floor mats
3Productivity
If wearable sensor systems are used for gait monitoring, then data capture during everyday activities is improved, but the system becomes intrusive and cannot be worn at all times
Solution Approach 1:
The patent replaces wearable mechanical sensors (accelerometers, gyroscopes, force sensors) with a non-contact electromagnetic radar system. This substitution eliminates the need for physical attachment to the patient, allowing continuous monitoring during all activities including bathing and sleeping, while maintaining the ability to capture gait data during everyday activities
4Loss of information
If traditional machine learning with separate feature extraction is used, then algorithmically describable features can be processed, but indescribable salient gait patterns cannot be effectively analyzed
Solution Approach 1:
The patent transforms the algorithmic approach by changing from traditional machine learning with separate feature extraction to deep learning with automated hierarchical feature learning. The deep neural networks automatically learn salient features directly from raw radar signals and spectrograms, capturing both algorithmically describable and indescribable gait patterns without manual feature engineering, thereby reducing information loss while managing complexity through end-to-end learning
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach significantly enhances the accuracy of gait estimation and Alzheimer's disease risk assessment, enabling the use of micro-Doppler radar as a pervasive indoor monitoring system for early-stage AD screening, providing a non-intrusive, cost-effective, and privacy-protected solution for continuous AD risk monitoring.
Implementation Method 1
MDR senses micro-motion-induced Doppler shifts and measures micro Doppler signatures (MDS)s in the joint time-frequency domain of human body parts
Implementation Method 2
interferometric technology to extend existing MDR as 'interferometric micro Doppler radar (IMDR)' system
Implementation Method 3
transmit an electromagnetic wave; receive a reflected electromagnetic wave
Data Source
AI summary
A system and method for quantifying Alzheimer's disease (AD) risk using one or more interferometric micro-Doppler radars (IMDRs) and deep learning artificial intelligence to distinguish between cognitively unimpaired individuals and persons with AD based on gait analysis. The system utilizes IMDR to capture signals from both radial and transversal movement in three-dimensional space to further increase the accuracy for human gait estimation. New deep learning technologies are designed to complement traditional machine learning involving separate feature extraction followed-up with classification to process radar signature from different views including side, front, depth, limbs, and whole body where some motion patterns are not easily describable. The disclosed cross-talk deep model is the first to apply deep learning to learn IMDR signatures from two perpendicular directions jointly from both healthy and unhealthy individuals. Decision fusion is used to integrate classification results from feature-based classifier and deep learning AI to reach optimal decision.