Camera-Based Patient Activity Classification in ICU Beds
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current monitoring technologies in intensive care units have limited ability to automatically and accurately classify patient activities, which is crucial for assessing health status, sedation depth, and mobility, and are often hindered by the need for frequent manual checks by nursing staff, leading to increased costs and potential inaccuracies due to external influences.
Innovation Solution
A computer-implemented method using camera-based systems to classify patient activities by distinguishing between intrinsic (patient-induced) and extrinsic (externally caused) activities through image data analysis, employing multiple sensors and deep learning techniques to provide robust detection and classification, and adaptively tracking zones of interest to reduce manual intervention.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If pressure sensors are attached to the patient's mattress, then patient activity can be monitored, but cleaning and hygiene become challenging and the detection range is static
Solution Approach 1:
The patent replaces contact-based mechanical pressure sensors with optical camera-based detection systems. This substitution eliminates the need for physical sensors on the mattress, making cleaning and hygiene much easier while maintaining the ability to monitor patient activity through image analysis of patient movements and positions.
2Reliability
If acceleration sensors are attached directly to the patient, then patient activity can be monitored, but the system cannot distinguish between intrinsic and extrinsic activities
Solution Approach 1:
The patent introduces camera-based observation as an intermediary between the patient and the monitoring system. By capturing images of the patient's overall movements and positions in the bed, the system can analyze patterns to distinguish between intrinsic activities (patient's own movements) and extrinsic activities (movements caused by external factors like bed adjustments or nursing care), thereby recovering the lost classification information.
3Measurement precision
If manual monitoring by nursing staff is performed, then patient activity can be assessed, but costs increase and manual checks are required at very short intervals
Solution Approach 1:
The patent implements an automated monitoring system that performs patient activity assessment independently without requiring continuous nursing staff intervention. The camera-based system continuously captures and analyzes patient movements, automatically generating activity assessments and alerts, thereby enabling the system to serve itself and eliminating the need for frequent manual checks while maintaining high measurement precision.
4Extent of automation
If camera-based systems are used to monitor patient activity, then automated detection is possible, but the system must distinguish between intrinsic and extrinsic activities to improve classification accuracy
Solution Approach 1:
The patent segments the monitoring task into distinct components: first detecting patient movements and positions from camera images, then classifying these movements as either intrinsic or extrinsic based on movement patterns, timing, and context. This segmentation of the classification process into manageable stages reduces the overall complexity while maintaining high automation capability.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Exemplary embodiments create a method, a device and a computer program for classifying activities of a patient (100) on the basis of image data of the patient (100) in a patient-supporting device (110). The method (10) comprises detecting (12) an activity of the patient (100) on the basis of the image data and determining classification information for the activity from the image data. The classification information comprises at least information about whether the detected activity of the patient (100) has been brought about actively by the patient (100) or passively by an external influence. The method (10) also comprises providing (16) information about the activity and the classification information.