Activity-Based Health Anomaly Model Updating for Chronic Changes
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing anomaly detection techniques in health monitoring systems fail to accurately detect small changes in physical conditions due to chronic changes in the condition of subjects, such as elderly individuals, leading to a decrease in detection performance.
Innovation Solution
A trained model generation method that updates the anomaly detection model using recent activity data to adapt to chronic changes in physical conditions, involving the generation of a second trained model when differences are detected between past and current activity data, ensuring the model reflects the current subject state.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If an anomaly detection model is trained using past activity data, then the model can detect anomalies based on historical patterns, but the detection precision decreases when chronic changes in physical condition occur
Solution Approach 1:
The patent implements dynamic model updating by comparing activity data across different time periods (first period, second period, and third period). When significant differences are detected between periods, the trained model is regenerated to adapt to chronic changes in the subject's physical condition, thereby maintaining high anomaly detection precision over time
Solution Approach 2:
The system incorporates feedback mechanisms by continuously monitoring activity data and comparing it across time periods. The comparison results feed into the model update decision process, where detected differences trigger model regeneration, creating a closed-loop system that adapts to changing conditions while maintaining detection reliability
2Measurement precision
If the anomaly detection model is updated frequently to adapt to chronic changes, then the detection precision improves, but the system complexity and computational resources increase
Solution Approach 1:
The patent changes the parameter of model update timing from continuous or fixed-interval updates to event-driven updates based on detected differences in activity data. The model is regenerated only when the comparison between activity data periods reveals significant changes, optimizing the balance between detection precision and system complexity
3Measurement precision
If the model is regenerated using recent activity data, then the detection accuracy for current state anomalies improves, but the processing time and computational cost increase
Solution Approach 1:
The system performs preliminary comparisons of activity data across different time periods before triggering model regeneration. By detecting differences in advance through systematic comparison of first, second, and third period data, the system prepares for model updates only when necessary, reducing unnecessary processing time and computational waste
Data Source
AI summary
A trained model generation method is a method for generating a trained model that uses, as an input, a feature based on activity data on a subject and outputs an anomaly score indicating a degree of an anomaly in a physical condition of the subject. A first trained model is a model that has been trained using first activity data on the subject. The first activity data is obtained during a first period. The trained model generation method includes: obtaining the first activity data on the subject and second activity data on the subject, the second activity data being obtained during a second period that is after the first period; and causing a second trained model to be generated using third activity data on the subject when a difference exists between the first activity data and the second activity data, the third activity data being obtained most recently.