CoughPhase-CLR: Designing an acoustics-informed foundation model for coughing sound classification

2026-06-19

AI Summary

Existing self-supervised methods overlook the physiological phase characteristics inherent in cough sounds, limiting their representation learning efficacy. This work proposes a cough-phase-aware self-supervised contrastive learning framework that, for the first time, incorporates acoustic-physiological phase information into pretraining. By constructing phase-aligned positive pairs and employing domain-specific data augmentation strategies, the approach overcomes the limitations of generic audio pretraining paradigms. Experimental results across five downstream tasks—including COVID-19 detection and COPD status classification—demonstrate substantial improvements over random cropping baselines, with a notable 57% unweighted average recall (UAR) achieved on COPD classification, underscoring both the method’s effectiveness and the inherent difficulty of the task.

Abstract

In this work, we introduce CoughPhase-CLR, a self-supervised learning framework designed to leverage the physiological phases of a cough for robust representation learning. Unlike generic contrastive frameworks, CoughPhase-CLR constructs positive pairs based on these specific acoustic phases. We pre-trained our model on approximately 40 hours of public cough audio and evaluated it across five downstream tasks, including COVID-19 detection, chronic obstructive pulmonary disease (COPD) state classification, and smoker status prediction. Our results demonstrate that cough-specific pre-training consistently outperforms standard random-cropping techniques when training on cough recordings. Additionally, we benchmarked a diverse set of state-of-the-art models on COPD state classification, highlighting the difficulty of this task. The best-performing models, pretrained on either general audio or respiratory sounds, achieved a UAR of 57\%, failing to outperform the state-of-the-art performance of 84\% UAR achieved using speech analysis.

Problem

• cough sound classification
• physiological phases
• COPD classification
• self-supervised learning
• acoustic representation

Innovation

• CoughPhase-CLR
• self-supervised learning
• acoustic phases
• cough sound classification
• contrastive learning