Estimating sub-epidermal moisture using spectroscopy and machine learning for early detection of pressure injury

Early detection of pressure injuries is crucial due to their negative impact on both patients and the healthcare system. This study employs spectroscopy and Partial Least Squares Regression (PLSR) to estimate skin moisture content as a biomarker for the early detection of pressure injuries. Moisture was measured at six points on each hand (one moisturized and one dry) using the MoistureMeterEpid, SEM Scanner, and two spectrometers (PIPA, VIAVI), with participants of varying skin tones. The data underwent preprocessing with Standard Normal Variate (SNV), and the PLSR model was trained to predict moisture levels from the spectrometer data using the reference measurements.