Institute of Fundamental Technological Research

Regarding sleep research, polysomnography (PSG) also called a sleep study, is a gold standard. It incorporates brain waves, the oxygen level in the blood, heart rate and breathing, and leg movement recordings. PSG is a complicated and expensive laboratory-based procedure, usually done in hospitals or special sleep center. In this study, an alternative technique for Sleep-Related Breathing Disorders (SRBD) based on selected cardiac and acoustic parameters and the Random Forest (RF) has been studied. A system dedicated to the detection of simultaneously acquired ECG and acoustic signals, which are collected during sleep at home environment is proposed. Results obtained indicate that classification and regression tree models such as RF are appropriate for the evaluation of sleep disorders like SRBD. The best identification of sleep irregularities at level 89.00 percent for the raw database was obtained. Thus, statistical predictive models allow identification of breathing events with high levels of sensitivity and specificity, providing an inexpensive and accurate diagnosis.

patient monitoring, random forest, disorders, biomarkers

The growing use of 3D-printed (additively manufactured) structural components implies the need to develop effective methods of damage assessment. This study focuses on guided wave propagation and its interaction with structural damage. The waves were excited using a laser scanning system which allows for easy excitation of the waves at various points at the surface. Also, the excitation is broadband, giving the ability to excite more guided wave modes at once. The combined laser scanning with a single piezoelectric measurement transducer takes advantage of reciprocity to reconstruct the full propagating wavefield. The investigated sample was printed from an aluminum alloy. The first set of measurements was realized for an intact (healthy) sample. Next, an artificial damage was introduced in order to study the wave interaction with it. Machine learning-based signal process algorithms were developed to analyze the wave interaction with the damaged plate. The obtained results show a good potential of guided wave-based techniques for the structural health monitoring of 3D-printed structures. © 2024, The Society for Experimental Mechanics, Inc.

3D printing; Guided waves; k-means clustering; Laser ultrasonics; Nondestructive evaluation; Principal component analysis