Fall Risk Prediction Using 3D Skeleton Reconstruction and ML
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current fall risk assessment methods, such as the Berg Balance Scale (BBS), are time-consuming, subjective, and rely on limited medical resources, making it difficult to efficiently identify and monitor older adults at high risk of falling, which is a significant public health concern due to the increasing elderly population.
Innovation Solution
A computer-implemented method using machine learning models that predicts fall probability by receiving and analyzing temporal data from sensors, reconstructing 3D scenes and skeletons, extracting spatio-temporal features, and introducing these features into a trained machine-learning model to predict fall risk, potentially reducing the number of BBS tasks required for assessment.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the Berg Balance Scale (BBS) is used for fall risk assessment, then the assessment is comprehensive and accurate, but it is time-consuming and requires limited medical professional resources
Solution Approach 1:
The patent replaces the mechanical/manual BBS assessment system with an automated computer vision system using sensors and machine learning algorithms. The system captures temporal data from multiple sensors, reconstructs 3D scenes and skeletons, extracts spatio-temporal features, and predicts fall risk automatically, eliminating the need for manual medical professional assessment while maintaining comprehensive evaluation capability
Solution Approach 2:
The patent creates a digital copy of the BBS assessment process through virtual 3D scene reconstruction and skeleton tracking. By capturing real-world movements through sensors and creating corresponding digital representations, the system replicates the comprehensive BBS evaluation methodology in a automated format that can be processed rapidly by machine learning models
2Measurement precision
If the Berg Balance Scale (BBS) is used for fall risk assessment, then the assessment is comprehensive and accurate, but it relies on limited medical professional resources
Solution Approach 1:
The patent substitutes the complex human expert assessment system with an automated computational system. Machine learning models trained on comprehensive features process sensor data to deliver accurate fall risk predictions without requiring medical professionals, thereby maintaining high assessment accuracy while eliminating dependency on limited human resources
Solution Approach 2:
The system enables self-service fall risk assessment where users can be evaluated without medical professional intervention. The automated machine learning pipeline independently performs data processing, feature extraction, and risk prediction, making comprehensive assessment accessible to broader populations without requiring scarce medical expertise
3Productivity
If fewer BBS tasks are used for assessment, then the assessment is faster and more efficient, but the quality and accuracy may be reduced
Solution Approach 1:
The patent transforms the assessment approach by changing from evaluating discrete BBS task scores to analyzing continuous spatio-temporal features of movement patterns. By extracting parameters such as velocity, acceleration, joint angles, and coordination metrics from sensor data, the system achieves comprehensive evaluation through fewer physical tasks while maintaining or improving accuracy through richer quantitative measurements
4Productivity
If automated machine learning assessment is implemented, then the screening capacity is expanded and efficiency is improved, but the system complexity increases
Solution Approach 1:
The patent divides the automated assessment system into distinct functional modules: data acquisition from multiple sensors, temporal 3D scene reconstruction, skeleton extraction, spatio-temporal feature engineering, and machine learning prediction. This segmentation allows each component to be optimized independently and facilitates deployment of screening capacity across multiple locations while managing system complexity through modular architecture
Data Source
AI summary
A computer-implemented method for predicting risk of fall of a user is disclosed. The method includes: receiving temporal data related to a skeleton of the user, from at least two sensors; reconstructing from the temporal data at least one of, a temporal 3D scene reconstruction, and a temporal three-dimensional (3D) skeleton reconstruction; extracting spatio-temporal features from the 3D scene reconstruction or the 3D skeleton reconstruction; introducing at least one spatio-temporal feature to a machine-learning (ML) model, wherein said ML model is trained to predict fall probability of a user based on said spatio-temporal feature; and predicting fall probability of the user based on an output of the ML model.