RAMoEA-QA: Hierarchical Specialization for Robust Respiratory Audio Question Answering

Researchers introduced RAMoEA-QA, a new AI system that uses hierarchical specialization to answer questions about respiratory audio recordings from mobile devices. The system employs a two-stage routing approach with Audio Mixture-of-Experts and Language Mixture-of-Adapters to handle diverse recording conditions and query types, achieving 0.72 test accuracy compared to 0.61-0.67 for existing baselines.

• →RAMoEA-QA addresses the challenge of analyzing heterogeneous respiratory audio recordings that vary across devices, environments, and acquisition protocols.
• →The system uses two-stage conditional specialization with Audio Mixture-of-Experts and Language Mixture-of-Adapters for improved performance.
• →The model achieved 0.72 in-domain test accuracy, significantly outperforming state-of-the-art baselines at 0.61-0.67.
• →The system demonstrates strong generalization capabilities for medical diagnosis under domain, modality, and task shifts.
• →This represents advancement in conversational AI for healthcare applications, particularly in respiratory care screening and monitoring.