Remote monitoring of respiratory diseases using AI that analyzes cough sounds and breathing patterns

Abstract

This study developed and evaluated "RespiroGuard", a remote monitoring system based on Artificial Intelligence (AI) that analyzes cough sounds and respiratory patterns for the management of chronic respiratory diseases. The mixed methodology included the development of deep learning algorithms and a prospective validation with COPD patients. The AI models achieved high performance in cough classification (AUC 0.98) and wheeze detection (AUC 0.94). In the clinical study, the system demonstrated significant predictive capacity for exacerbations (AUC 0.82), with a Negative Predictive Value of 96.3%, highlighting its utility as a "reassurance" tool. Qualitative analysis revealed that patients valued the safety and empowerment provided by objective data, but also experienced ambivalence towards constant monitoring, sometimes perceiving it as intrusive. It is concluded that the technology is clinically promising, but its successful implementation requires ethical design that prioritizes user control and seamless integration into clinical workflows, balancing technical accuracy with the human patient experience