Abstract Introduction Inertial measurement units (IMUs) are increasingly integrated into home-care and wearable devices. When placed on the fingertip, we observed that IMU accelerometer signals capture subtle peripheral motion associated with respiratory effort. This modality may provide complementary information when traditional respiratory channels are unreliable or unavailable, offering potential value for downstream applications such as obstructive sleep apnea (OSA) screening. Methods Tri-axial IMU accelerometer signals were preprocessed by detrending and bandpass filtering to isolate physiologic frequency bands. We modeled each channel as an adaptive non-harmonic oscillation and analyzed it using the synchrosqueezing transform (SST), a nonlinear time–frequency analysis method. The respiratory phase was obtained via ridge extraction of the putative respiratory component encoded in the IMU signal. Phase-based time unwarping was then applied to stabilize nonstationary oscillations and concentrate spectral energy. A Respiratory Quality Index (RQI) was computed from the Fourier transform of the unwarped signal, defined as the ratio of harmonic respiratory energy to total cardiopulmonary-band energy. Epochs exceeding a predefined RQI threshold were retained for waveform extraction using shape-adaptive mode decomposition (SAMD), a recently developed signal decomposition algorithm capable of recovering nonstationary physiological oscillations. Estimated fingertip-derived respiratory waveforms were compared with reference airflow measurements. Results A total of 124 full-night IMU recordings collected using the TipTraQ device were analyzed. A substantial proportion of epochs exhibited high RQI, indicating consistently usable signal quality for respiratory estimation. In high-quality segments, the reconstructed fingertip respiratory waveform showed strong temporal concordance with reference airflow, including high waveform correlation and low respiratory-rate error. Conclusion This study demonstrates the feasibility of deriving robust respiratory waveforms from fingertip IMU recordings obtained from wearable devices. The combination of RQI-based quality assessment and SAMD-based waveform reconstruction enables reliable extraction of respiratory dynamics from a single IMU sensor. These findings highlight the potential of fingertip-only IMU monitoring as a complementary or alternative respiratory channel for home-based sleep assessment and OSA screening. Support (if any)
Liu et al. (Fri,) studied this question.