Detecting Patient Position Using Bed-Reaction Forces for Pressure Injury Prevention and Management

Area of Science:

• Biomedical Engineering
• Clinical Nursing
• Rehabilitation Technology

Background:

• Regular repositioning is crucial for preventing and treating pressure injuries (PIs) in at-risk individuals.
• Non-contact position detection systems using bed load cells offer a promising approach for monitoring patient position.
• Previous research indicated a need for higher precision in position detection to ensure off-loading of high-risk bony prominences.

Purpose of the Study:

• To evaluate the impact of increased position categorization precision on the performance of a non-contact bed sensor system.
• To determine if higher precision in predicting patient position improves the system's ability to identify off-loading of pelvic bony prominences.
• To assess the relationship between position category bin size and the prediction F1 score of the developed system.

Main Methods:

• Utilized data from 18 participants, collected via load cells under bed legs and a pelvis-mounted inertial measurement unit.
• Trained classifiers to predict transverse pelvic angle using three distinct position bin sizes: 45°, ~30°, and 15°.
• Employed a leave-one-participant-out cross-validation approach to rigorously evaluate classifier performance for each bin size.

Main Results:

• The prediction F1 score demonstrated a decline as the precision of position categorization was increased.
• Classifiers trained with larger position bins (lower precision) achieved higher F1 scores compared to those trained with smaller bins (higher precision).
• This suggests that the system's accuracy in predicting patient position is inversely related to the desired level of positional detail.

Conclusions:

• Increasing the precision of position detection in the developed non-contact system led to a decrease in prediction accuracy (F1 score).
• The findings suggest that simpler, less precise position classifications may be more reliable for this specific sensor technology.
• Further research is needed to optimize sensor algorithms or explore alternative methods for achieving accurate, high-precision position detection for pressure injury prevention.