JMIR Publications

ABSTRACT

Background:

Asthma remains one of the most serious chronic diseases of childhood. Individuals with severe asthma experience sudden episodes of breathlessness due to acute airflow obstruction, leading to recurrent Pediatric Intensive Care Unit (PICU) admissions that often result in mechanical ventilation and even death. Existing clinical assessments lack temporal resolution to effectively track the rapidly changing physiology.

Objective:

This study aimed to evaluate the feasibility of quantifying respiratory health during acute asthma in children using wearable multimodal sensing.

Methods:

Wearable-based Impedance Pneumography (IP) and multi-channel breath sounds were measured on 17 children admitted to the PICU with an acute asthma attack, and on 9 healthy controls. Multimodal measurements were obtained throughout hospitalization, from PICU admission (T1) to discharge (T2). Statistical and clustering analyses were performed to identify trends in IP- and LS-derived respiratory markers from T1 to T2 and were compared against matched controls (CTL). Leveraging the breathing context provided by the IP signal, spectral integrated intensities (SI) acoustic features were computed in four different sub-bands for the inspiration (I) and expiration (E) phases.

Results:

Within the patient group (n=13), we found that Respiratory Rate (RR) decreased (p=0.017), whereas expiration time (Te, p=0.021) and inspiration time (Ti, p=0.017) lengthened. Meanwhile, the SIs for the lowest sub-band (100-300Hz) decreased for both I (p<0.01) and E phases, while they increased for the highest sub-band (800-1000Hz) decreased for both I and E phases (p<0.01). Significant differences also existed across the multimodal respiratory features between T1 (n=10) and CTL (n=9), T2 (n=7) and CTL of a matched cohort. We found that all features trended toward normalcy for the patients.

Conclusions:

We demonstrated the feasibility of quantifying and tracking respiratory health in children with acute asthma using wearable multimodal sensing, specifically with the fusion of IP and LS-derived markers. Significance: This work presents the first demonstration of wearable-based sensing for quantifying the physiologic manifestations of a life-threatening asthma attack from peak to resolution of symptoms, thereby motivating further research aimed at personalizing care and improving patient outcomes.