A Wearable Multimodal Sensing System for Tracking Changes in Pulmonary Fluid Status, Lung Sounds, and Respiratory Markers

Abstract

Heart failure (HF) exacerbations, characterized by pulmonary congestion and breathlessness, require frequent hospitalizations, often resulting in poor outcomes. Current methods for tracking lung fluid and respiratory distress are unable to produce continuous, holistic measures of cardiopulmonary health. We present a multimodal sensing system that captures bioimpedance spectroscopy (BIS), multi-channel lung sounds from four contact microphones, multi-frequency impedance pneumography (IP), temperature, and kinematics to track changes in cardiopulmonary status. We first validated the system on healthy subjects (n = 10) and then conducted a feasibility study on patients (n = 14) with HF in clinical settings. Three measurements were taken throughout the course of hospitalization, and parameters relevant to lung fluid status-the ratio of the resistances at 5 kHz to those at 150 kHz (K)-and respiratory timings (e.g., respiratory rate) were extracted. We found a statistically significant increase in K (p < 0.05) from admission to discharge and observed respiratory timings in physiologically plausible ranges. The IP-derived respiratory signals and lung sounds were sensitive enough to detect abnormal respiratory patterns (Cheyne-Stokes) and inspiratory crackles from patient recordings, respectively. We demonstrated that the proposed system is suitable for detecting changes in pulmonary fluid status and capturing high-quality respiratory signals and lung sounds in a clinical setting.

Keywords: bioimpedance spectroscopy; cardiorespiratory monitoring; fluid status; heart failure; impedance pneumography; lung sounds; sensor fusion; wearable sensing.

Figure 1

(a) Overview of the redesigned system for respiratory monitoring. A central circuit housing connects to each sensor independently and can be attached to the arm via Velcro strap. The semi-transparent material of the box provides visual feedback through LEDs. The system contains four audio channels, 2 IMUs (1 reference), 2 temperature sensors (1 reference), and 2 pairs of EBI electrode wires. (b) Central circuit housing hosting the audio and main boards PCBs, 2 500 mAh batteries, and mechanical switches to initiate/stop recordings and control the operation mode. All connectors enter the box from the same side via right-angle connectors. (c) Custom 3D-printed microphone case to provide constant backing force. The contact microphones (BU-23173-000, Knowles Electronics LLC., Itasca, IL, USA) were professionally overmolded in a 77 A durometer silicone. (d) Placement of multimodal sensors utilized in this work. (e) Exemplary recording from a selected subset of sensors during a deep breathing maneuver.

Figure 2

Multi-frequency impedance pneumography (IP) signal processing pipeline. After filtering the signals, breaths are detected, and their signal quality is assessed. This assessment employs overlapping windows to enable breath-by-breath evaluations with the SQI published in (Charlton et al., 2021) at its core. A final stage of plausibility assessment ensures that only breaths yielding physiologically plausible respiratory rates (RR) are deemed as good quality. These good breaths are then used to extract amplitude (Rpki/e) and timing features (Ti/e, RR). Outliers are finally removed if any of the features lay outside ±4 median absolute deviations (MAD) from the overall median.

Figure 3

Validation against the Eko CORE digital stethoscope. The mean and standard deviations for both spectra are plotted. The data were obtained from 10 healthy volunteers with both sensors close to each other and over the posterior left lower chest quadrant. Both systems recorded simultaneously while the participants took deep breaths over a 30–s period in sitting position.

Figure 4

Results from proof-of-concept clinical recordings. (a) Differences between the admission and discharge groups for K (the ratio of the resistances at 5 kHz and 150 kHz) showing a statistically significant increase (p < 0.001, Wilcoxon signed-rank) from K = 1.27 ± 0.12 to K = 1.32 ± 0.15. This statistically significant increase in K indicates the reduction of pulmonary fluid or its redistribution into the appropriate intracellular compartments. (b) Differences in the mean RR from admission to discharge groups showing a slight decrease from 23.12 ± 5.53 bpm to 22.73 ± 6.85 bpm, not statistically significant. (c) Differences in the mean Te:Ti ratio showing a slight increase from 1.10 ± 0.27 to 1.23 ± 0.32, not statistically significant. * denotes a p-value lower than 0.05 and was considered statistically significant for this work.

Figure 5

Abnormal breathing pattern finding from the proof-of-concept clinical recordings. (a) Segment of multimodal data obtained from patient 13, in which they were breathing following the CSR pattern. (b) Seven–second segment from the first CSR event in (a). (c) Placement of sensors used in the recordings. (d) Mean BIS curve for this patient.

Figure 6

Time and time–frequency visualization of the first CSR event in Figure 5 for Ch1 (anterior right upper chest quadrant). The time–frequency visualization was obtained through STFT analysis using 300 ms windows and 95% overlap.

Figure 7

Finding of inspiratory (INS) crackles from Ch3 (posterior left lower chest) contextualized by the concurrent IP-derived respiratory signal (IP_{100 kHz}) (top). Time (middle) and time–frequency (bottom) representations of the recorded sounds are shown. The time–frequency visualization was obtained through STFT analysis using 250 ms windows and 95% overlap.