Toward mitigating pressure injuries: Detecting patient orientation from vertical bed reaction forces

Area of Science:

• Biomedical Engineering
• Machine Learning in Healthcare
• Patient Monitoring Systems

Background:

• Prolonged immobility can cause pressure injuries.
• Adhering to repositioning schedules is difficult for caregivers and patients.
• A monitoring device could reduce pressure injury incidence and severity.

Purpose of the Study:

• To develop a method for detecting patient orientation in bed.
• To utilize data from load cells under hospital beds.
• To create an alert system for caregivers to prevent pressure injuries.

Main Methods:

• Twenty individuals were placed in supine, left side-lying, or right side-lying positions.
• Force data was used to extract breathing pattern characteristics.
• Machine learning models (Logistic Regression, Feed Forward Neural Network) were trained and validated.

Main Results:

• The Feed Forward Neural Network achieved 94.2% accuracy in predicting patient orientation.
• Breathing pattern characteristics were key features for classification.
• The system demonstrated high accuracy in identifying different bed positions.

Conclusions:

• The non-invasive system shows high accuracy in detecting patient position.
• This technology has potential for developing pressure injury prevention tools.
• The algorithm could significantly improve patient care by enabling timely repositioning.