A sensitivity analysis to evaluate the performance of temporal pressure - related parameters in detecting changes in supine postures.

Pressure mapping systems have been traditionally used to assess load distributions in individuals at risk of pressure ulcers. Recently, the technology has been adapted to monitor pressures over prolonged periods. The present study aims to investigate the predictive ability of selected biomechanical parameters estimated from pressure distributions for detecting postural changes in lying. Healthy participants (n = 11) were recruited and positioned in different lying postures, by utilizing the head of bed (HOB) angle and an automated tilting system to achieve evoked movements in the sagittal and transverse planes, respectively. Measurements included continuous monitoring of interface pressures and accelerations from the trunk and waist. Selected interface pressure parameters included; centre of pressure, contact area and pressure gradient. A threshold range for all parameters was established and Receiver Operating Characteristic (ROC) curves presented to determine the sensitivity and specificity for detecting postural changes. Temporal trends in the data revealed significant variance in the signal perturbations during each evoked postural change. Indeed, sensitivity and specificity were influenced by the specific threshold values and the relative position of the individual. As an example, sensitivity of some parameters exhibited a compromised trend at higher HOB angles, with low corresponding area under the ROC curve. By contrast, contact area provided the highest values, with 7/12 signals achieve AUC >0.5. This corresponded with actimetry signals, which provided high discrimination between postures. Parameters estimated from a commercial pressure monitoring can have the potential to detect postural changes. Further research is required to convert the data into meaningful clinical information, to inform patient repositioning strategies.