The use of entropy to assess sleep-disordered breathing in asthmatic patients

Abstract

Background: Asthma is a chronic respiratory disease affecting 300 million people worldwide, creating social, financial, and emotional burdens. When combined with obstructive sleep apnoea (OSA), the impact on individuals and healthcare systems is greater. People with uncontrolled asthma are twice as likely to develop sleep-disordered complications. Conventional diagnostic methods such as the Apnoea–Hypopnea Index (AHI), Oxygen Desaturation Index (ODI), mean heart rate (HR), and oxygen saturation (SpO₂), lack sensitivity and may miss subtle irregularities. Nonlinear approaches, such as sample entropy (SampEn), offer potential advantages by capturing hidden dynamics in HR variability (HRV) and SpO₂ signals.

Aim: This study investigated whether entropy measures derived from HRV and SpO₂ could differentiate asthmatic patients (with/without sleep-disordered breathing, SDB) from controls.

Methods: A prospective observational study was conducted at the Royal Free Hospital (May–July 2025). Twenty participants were recruited (10 asthma, 10 controls). Overnight monitoring was performed using the Embletta MPR Gold (Natus Medical, USA) to record HR and SpO₂. SampEn was calculated in MATLAB R2024a. Group comparisons were assessed using ANOVA and Tukey post hoc tests (p < 0.05).

Results: Mean SpO₂ did not differ significantly between groups (p = 0.18). Mean HR was higher in asthma than controls (p = 0.04) but did not distinguish asthma subgroups. Conventional indices (AHI, ODI) increased in asthma but lacked sensitivity. SampEn HRV showed no significant differences (p = 0.979), whereas SampEn SpO₂ was significantly higher in the asthma group (p = 0.001) and effectively differentiated asthma+SDB from asthma−SDB.

Conclusion: SampEn SpO₂ emerged as a sensitive marker of irregular nocturnal respiratory dynamics in asthma, offering greater discriminatory power than conventional indices. Despite limitations such as small sample size, findings highlight entropy’s potential for clinical assessment of asthma with SDB. Future work should involve larger cohorts, multi-night monitoring, and integration with machine learning.