Kinematic signature of high-risk labored breathing revealed by novel signal analysis

Abstract

AbstractBreathing patterns (respiratory kinematics) contain vital prognostic information. They report on a dimension of physiology that is not captured by conventional vital signs. But for an informative physiomarker to become clinically valuable, it must be measureable with ease, accuracy, and reproducibility. We sought to enable the quantitative characterization of respiratory kinematics at the bedside. Using inertial sensors, we analyzed upper rib, lower rib, and abdominal motion of 108 patients with respiratory symptoms during a hospital encounter (582 two-minute recordings). We measured the average respiratory rate and 33 other signal characteristics that had an explainable correspondence with clinically significant breathing patterns. K-means clustering revealed that the respiratory kinematic information was optimally represented by adding 3 novel measures to the average respiratory rate. We selected measures representing respiratory rate variability, respiratory alternans (rib-predominant breaths alternating with abdomen-predominant ones), and recruitment of accessory muscles (increased upper rib excursion). Latent profile analysis of these measures revealed a phenotype consistent with labored breathing. Poisson regression showed that the rate at which a patient’s recordings exhibited the labored breathing phenotype was significantly associated (p<0.01) with the severity of illness (discharge home v/s acute-care hospitalization v/s critical-care hospitalization). Notably, labored breathing was frequently detectable (21%) when the respiratory rate was normal, and it improved discrimination for critical illness. These findings validate the feasibility of respiratory kinematic phenotyping in routine healthcare settings, and demonstrate its clinical value. Further research into respiratory kinematic characteristics may reveal novel pathophysiologic mechanisms, advance the efficacy of predictive analytics, and enhance patient safety.