Bed Exit Prediction Using Multi-Parameter Force Signal Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing bed exit prediction systems often produce false alarms, placing an undue burden on caregivers and requiring reduction in false alarm likelihood while accurately predicting actual bed exits.
Innovation Solution
A system using a processor, memory, and force sensors to analyze force signal properties such as mean frequency, median frequency, peak frequency, standard deviation of frequency, and mean energy, generating a notification only if all criteria are met, thereby reducing false alarms and improving prediction accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If a bed exit prediction system uses simple alarm triggers, then the system is easy to operate, but false alarms increase placing an undue burden on caregivers
Solution Approach 1:
The system changes the parameter of alarm triggering from simple binary thresholds to a comprehensive analysis of multiple signal properties (mean frequency, median frequency, peak frequency, standard deviation of frequency, and mean energy). This multi-parameter approach transforms the reliability of the system by reducing false alarms while maintaining ease of operation through automated analysis.
Solution Approach 2:
The system moves from one-dimensional alarm triggering to five-dimensional signal analysis by evaluating mean frequency, median frequency, peak frequency, standard deviation of frequency, and mean energy simultaneously. This dimensional expansion allows the system to distinguish true exit events from false alarm conditions more effectively.
2Reliability
If the system analyzes multiple signal properties to reduce false alarms, then prediction accuracy improves, but device complexity increases
Solution Approach 1:
The system applies multi-functionality by using existing force sensors (load cells) for their primary weight measurement function while simultaneously analyzing their output signals for exit prediction. The same sensor infrastructure serves dual purposes: monitoring occupant weight and detecting exit events through frequency and energy analysis, thereby improving prediction accuracy without proportionally increasing device complexity.
Solution Approach 2:
The system implements self-service by automatically analyzing multiple signal properties and making prediction decisions without requiring manual configuration or intervention. The automated processing of mean frequency, median frequency, peak frequency, standard deviation, and mean energy criteria reduces the operational burden on caregivers while maintaining high prediction accuracy.
3Device complexity
If existing force sensors are used for both weight measurement and exit prediction, then cost and complexity are reduced, but measurement precision for exit prediction may be compromised
Solution Approach 1:
The system changes the measurement approach by transforming the static weight data from force sensors into dynamic frequency and energy characteristics. By analyzing mean frequency, median frequency, peak frequency, standard deviation of frequency, and mean energy of the force signal, the system extracts rich exit prediction information from existing sensors, maintaining measurement precision without requiring additional specialized sensors.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A system for predicting exit from an occupant support, includes a processor, a memory in communication with the processor, and a frame having at least one force sensor which outputs a force signal in response to force exerted thereon. The system also includes machine readable instructions stored in the memory which cause the system to perform at least the following actions when executed by the processor: 1) determine a property of the signal during an interval of time, 2) classify the property as suggesting an exit event or as not suggesting an exit event, and 3) if the property is classified as suggesting an exit event, generate a notification thereof.