AI-enhanced chest physiotherapy for secretion clearance

Bronchiectasis and other chronic respiratory conditions associated with excessive airway secretions require regular chest physiotherapy to mobilize mucus, improve expectoration, reduce respiratory symptoms, and support functional capacity. Conventional airway clearance techniques (ACTs), including active cycle of breathing technique (ACBT), autogenic drainage (AD), postural drainage, positive expiratory pressure (PEP), oscillating positive expiratory pressure (OPEP), and manual chest physiotherapy, are widely used, however, selecting the most appropriate technique, determining the optimal treatment position, and evaluating the effectiveness of each session remain dependent largely on clinical assessment and patient-reported symptoms. Recent advances in digital auscultation, wearable acoustic sensors, signal processing, and artificial intelligence (AI) offer opportunities to improve respiratory physiotherapy through objective monitoring of lung sounds and individualized treatment recommendations. Acoustic sensor arrays can record respiratory sounds from multiple thoracic locations, potentially identifying regional differences in wheezes, crackles, rhonchi, airflow intensity, and other acoustic characteristics associated with altered airway mechanics and secretion movement. Machine learning (ML) algorithms can analyze these signals, identify patterns, and support clinical decisions regarding the selection and modification of airway clearance techniques. The proposed application is particularly relevant to bronchiectasis, in which mucus retention, impaired mucociliary clearance, airway inflammation, and recurrent infection contribute to persistent respiratory symptoms. This review examines the physiological basis of secretion clearance, the potential role of acoustic sensor arrays, ML-based respiratory sound classification, and the integration of AI into ACBT, autogenic drainage, positioning, and other physiotherapy interventions. Particular attention is given to three clinically meaningful outcomes: sputum volume cleared, localized chest expansion changes, and patient self-efficacy in secretion clearance. Although preliminary research demonstrates that automated respiratory sound classification is technically feasible, current evidence does not establish that acoustic AI can reliably identify mucus location or independently select the most effective ACT in bronchiectasis. Therefore, AI-enhanced chest physiotherapy should currently be considered a proposed clinical decision-support approach requiring prospective validation, rather than an established autonomous treatment system.