Machine learning methods to evaluate fluctuation in oxygenation and peripheral tissue temperature, in sacral and trochanteric area under pressure in healthy subjects and institutionalized patients. A prospective non-randomized trial

Abstract

peripheral oxygenation and local temperature of sacral and trochanteric areas exposed to pressure, comparing a healthy population with one at risk of impaired skin integrity. Methods: The study protocol is registered in ClinicalTrials.gov (NCT02736838). Non-randomised experimental study in two phases. Volunteers were recruited at the University and nursing homes from Nov 2017 to June 2020. The parameters were measured in the sacral and trochanteric areas under pressure using a laser Doppler monitor and near-infrared spectroscopy over 2 h. Body positions varied by area: supine for the sacral region and lateral supine for the trochanteric region. Results: Comparative analysis between both groups of the sacral area showed significant fluctuations in perfusion (p = .035) and local temperature (p = .013). In the trochanteric area, the comparative analysis showed sig nificant fluctuations in perfusion (p = .044), local temperature (p > .001), and oxygenation (p = .049). Nonlinear patterns were identified in the variables examined in both anatomical areas. This prompted the adoption of an algorithm based on the Random Forest Regressor technique to predict the parameters associated with the development of pressure injuries. Conclusions: Individuals at risk of developing pressure injuries show a more limited compensatory physiological response compared to healthy individuals in terms of blood flow, oxygenation, and tissue temperature in pressure areas. Notably, at the 90-min, a turning point is observed, marking a critical threshold from which these pa rameters begin to decline.