Research on inspiratory muscle training action recognition based on passive RFID tags and hybrid deep learning.

The article focuses on the development and evaluation of a novel respiratory status sensing system for chronic obstructive pulmonary disease (COPD) rehabilitation, utilizing passive radio frequency identification (RFID) tags integrated into smart sensing clothing combined with a hybrid deep learning model. This system employs a CNN-LSTM-AM (convolutional neural network–long short-term memory–attention mechanism) architecture to non-invasively recognize inspiratory muscle training actions, achieving classification accuracies of 99.17% for abdominal breathing pattern recognition and 95.56% for identifying breathing stages (inhalation, breath holding, exhalation). The approach leverages passive RFID technology for remote, real-time monitoring of respiratory movements, aiming to enhance home-based COPD rehabilitation by providing accurate assessment and feedback without the need for active sensors or cameras. Limitations include high site and sampling rate requirements and system cost, with future work suggested to optimize these aspects and explore alternative attention mechanisms to improve clinical applicability.