Multimodal Gait Analysis System Using Sensor Fusion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current gait and posture analysis technologies rely on single modalities, failing to incorporate a wide range of data needed to create a patient's unique gait profile, and are inadequate for providing diagnoses or corrective measures for gait abnormalities and posture issues.
Innovation Solution
A system utilizing multimodal machine learning to classify gait, posture, proprioception, and kinesthesis by processing data on center of pressure (COP), center of gravity (COG), and posture, incorporating various data sources such as sensors, video, and clinical data to provide accurate diagnoses and corrective measures.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If single modality sensors are used for gait analysis, then device complexity is reduced, but measurement precision and diagnostic accuracy deteriorate
Solution Approach 1:
The patent combines multiple sensing modalities (accelerometers, gyroscopes, magnetometers, barometers, cameras, microphones) into an integrated gait analysis system. This merging of sensors allows the system to capture comprehensive gait data from different physical domains, thereby improving measurement precision and diagnostic accuracy without requiring separate single-modality systems.
Solution Approach 2:
The patent creates a universal gait analysis platform that can perform multiple functions: detecting gait parameters, classifying gait abnormalities, monitoring patient progress, and providing diagnostic information. This multi-functional system replaces the need for multiple specialized single-modality devices, achieving both comprehensive measurement and system integration.
2Measurement precision
If comprehensive multimodal data is collected, then diagnostic accuracy improves, but data processing complexity increases
Solution Approach 1:
The patent performs preliminary processing of multimodal data by extracting gait-specific features and parameters before final analysis. This preliminary action involves preprocessing sensor signals, filtering noise, and extracting relevant gait characteristics, which simplifies subsequent diagnostic processing while maintaining high diagnostic accuracy.
Solution Approach 2:
The patent introduces intermediary processing layers that translate raw multimodal sensor data into meaningful gait parameters and features. These intermediary representations serve as a bridge between complex raw data and final diagnostic conclusions, reducing processing complexity while preserving diagnostic information.
Data Source
AI summary
A method comprising: receiving, with respect to each of a plurality of subjects, data comprising at least one of: (i) a data representing center of pressure (COP) of a subject during at least on gait phase; (ii) a data representing center of gravity (COG) of the subject during at least on gait phase; and; (iii) a data representing posture of the subject during at least on gait phase; processing the data to extract a plurality of features representing the data, at a training stage, training a machine learning model on a training set comprising all of the plurality of features; and at an inference stage, applying the trained machine learning model to a target set of the features obtained for ma target subject, to classify posture, proprioception and/or kinesthesis of the target subject.