Patient Movement Detection Using Feature Point Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing safe nursing systems fail to accurately detect patient movements, particularly finer movements and those indicating potential falls from a bed, due to limitations in sensor placement and image capture methods, leading to errors and interference with caregiver work.
Innovation Solution
A safe nursing system with a control unit and image capturing unit positioned to capture images from the side of the bed, using feature points and predetermined indices to determine patient movements by calculating coordinates and movement directions, setting conditions based on bed borders, and outputting alerts when specific thresholds are exceeded.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If an image capturing unit is positioned to capture images from the side of the bed, then the detection of patient movements toward the side of the bed is improved, but the camera may interfere with nurses' work and increase errors due to nurses entering the captured image
Solution Approach 1:
The patent positions the image capturing unit to capture images from above the bed rather than from the side, changing the dimensional perspective of monitoring. This overhead positioning allows the camera to monitor the entire bed surface and patient movements without being in the nurse's working space, thus resolving the contradiction between detection accuracy and interference with caregiver work
Solution Approach 2:
Instead of placing the camera at nurse level (side view), the patent inverts the positioning to an overhead view. This inversion allows the camera to monitor patient movements from a perspective that does not obstruct nurse access to the patient, eliminating the interference problem while maintaining detection capability
2Productivity
If rising behavior is determined based on the size of the patient's image region, then the system can detect patient movement, but errors occur depending on the manner of rising, body shape differences, and whether the patient enters the monitoring region
Solution Approach 1:
The patent divides the patient's body into multiple feature points rather than treating the patient as a single image region. By tracking the positions and movements of multiple discrete feature points (head, shoulders, hips, knees, ankles), the system can accurately determine patient movements regardless of body shape variations or the specific manner of movement, eliminating the errors associated with size-based detection
Solution Approach 2:
The patent changes the detection parameter from image region size to feature point coordinates and movement vectors. By calculating the positional changes of multiple feature points and determining movement direction and magnitude based on coordinate differences, the system achieves accurate movement detection that is independent of patient body shape or the specific manner of movement
3Reliability
If sensors are placed to monitor patient movement, then fall prevention can be achieved, but finer movements (coughing, rolling over) cannot be recognized and notifications are made without proper determination or after the fall has occurred
Solution Approach 1:
The patent implements preliminary detection by continuously monitoring the positions of multiple feature points and calculating movement vectors in real-time. By detecting subtle changes in feature point coordinates before they constitute a full movement action, the system can identify finer movements such as coughing and rolling over early, enabling timely notifications before falls occur
Solution Approach 2:
The patent employs dynamic monitoring by tracking the temporal changes in feature point positions and calculating movement vectors that reflect the speed and direction of patient movement. This dynamic approach allows the system to distinguish between different types of movements (subtle vs. significant) and provide appropriate notifications based on the actual movement state, improving both detection precision and fall prevention reliability
Data Source
AI summary
A safe nursing system that monitors movement of a patient on a bed includes a control unit and an image capturing unit disposed in a position from which an image including a side of a top surface of the bed can be captured. The control unit obtains image data captured by the image capturing unit; specifies feature points in an image of the patient in the obtained image data, and calculates a value of a predetermined index of the feature points used to determine a predetermined movement in the patient; sets a predetermined condition to determine the predetermined movement based on a region of the bed in the obtained image data; determines the predetermined movement based on the calculated value and the set predetermined condition; and outputs information indicating that the patient has performed the predetermined movement under the condition that the predetermined movement has been determined to have been performed.