Figure 5.

Strategy for separation of respiratory signals when two persons share the bed. (a) Cartoon illustrating how two persons sharing a bed are modeled as two respiratory point sources and raw signals from the 4 legs of the bed beneath two sleeping individuals. Inset shows 2 sensors each predominantly measuring one person with contaminating signal from the second person. (b) Demixed signals for person 1 (blue) and person 2 (pink) derived from the raw signals in A. (c) Validation experiment comparing demixed BedScales signals (blue and pink) with the corresponding ground truth chest belt signals (black). (d) Bar plot comparing peak location errors across both subjects between the demixed signal and the “correct chest belt “versus the error between the demixed signal and the “wrong chest belt.” (e,f) Bar plot quantifying the error between the BedScales separated signal from subject A (e) or B (f) and the chest belt on subject A (A-CB, left) or subject B (B-CB, right). Data are shown as mean ± standard deviation. ****P < 0.0001, Mann–Whitney test.