Back

Non-contact and non-constraining monitoring of the respiratory rate including sleep disordered breathing using ultra-wideband radar

Chin, K.; Okumura, S.; Endo, D.; Nagata, K.; Ito, T.; Murase, K.; Sunadome, H.; Hoshi, M.; Hiranuma, H.; Kozu, Y.; Sato, S.; Hirai, T.; Gon, Y.; Sakamoto, T.; Taki, H.; Akahoshi, T.

2024-07-09 respiratory medicine
10.1101/2024.07.08.24310110 medRxiv
Show abstract

BackgroundThe respiratory rate (RR) is a critical vital sign controlled by not only metabolic factors but behavior while awake. The prevalence of obstructive sleep apnea (OSA) is remarkably high. Therefore, a non-constraining and non-contact respiratory monitor to measure the RR both while awake and asleep is preferable. MethodsA millimeter wave radar (MWR) device was used for RR monitoring of participants suspected of OSA while awake (supine before and after polysomnography [PSG], sitting, and positioned on both sides) and asleep. Apnea and hypopnea during 1 hour of sleep (AHI) by PSG was compared with 1 hour of respiratory events (REI) judged by MWR. Portable monitoring (PM) and percutaneous arterial O2 (SpO2) monitoring were done simultaneously. ResultsCorrelations between the RR by MWR and respiratory inductance plethysmography at PSG while awake in every measured position including the supine position before and after PSG were very high (r=0.92[~]0.99) (n=60). The REI by MWR were significantly correlated with AHI determined by PSG, PM, or SpO2 monitoring (p<0.001). Brand-Altman plot showed that the MWR for AHI monitoring was acceptable. Predicted AHI by MWR relative to PSG was almost the same as with PM or SpO2 monitoring. ConclusionsThe developed MWR respiratory monitor was useful during wakefulness and sleep, including detection of apnea and hypopnea. This system can be useful in multiple medical settings such as critical care with and without sleep apnea, pandemic infection, elder care at home, etc. Trial registration number: UMIN000045833 (http://www.umin/ac.jp)

Matching journals

The top 5 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.