Patient–ventilator interaction plays a critical role in assisted respiratory therapy, where synchronization between the subject’s natural breathing cycle and the mechanical airflow delivered by the ventilatory system determines treatment effectiveness and safety. Mismatch events, commonly referred to as patient–ventilator asynchrony, can lead to increased work of breathing, patient discomfort, prolonged ventilation duration, and adverse clinical outcomes. Existing detection approaches rely on manual waveform inspection, rule-based algorithms, or machine learning techniques; however, these methods often struggle with pseudo-periodic respiratory signals, noise contamination, and variability in breathing patterns. To address these limitations, this study proposes a respiration-phase guided partitioning method designed to improve the identification of mismatch events between assisted airflow systems and subjects through precise segmentation of breathing cycles and phase-aware analysis.

The proposed method introduces a structured partitioning framework that separates respiratory signals into physiologically meaningful phases, including inspiration, expiration, and transition intervals, allowing mismatch detection to be performed within phase-specific contexts rather than on entire signals. The approach integrates pseudo-period extraction techniques, dynamic time alignment, and feature-based classification to improve robustness against waveform variability and irregular breathing behavior. By incorporating phase-guided segmentation, the method enhances sensitivity to subtle timing inconsistencies that traditional global analysis methods fail to detect.

A theoretical foundation based on time-series segmentation, pseudo-periodic signal processing, and machine learning-based classification is developed to support the proposed framework. The method is evaluated using representative respiratory waveform scenarios, including normal synchronization, delayed triggering, premature cycling, and flow mismatch conditions. Experimental analysis demonstrates that respiration-phase guided partitioning enables more reliable detection of mismatch events compared with conventional waveform-level detection approaches.

The findings indicate that phase-aware analysis provides improved interpretability, higher detection accuracy, and better adaptability to variable breathing patterns. The proposed framework contributes to the development of intelligent monitoring systems for assisted airflow devices and supports the advancement of automated synchronization assessment in respiratory care. This work establishes a methodological basis for future research on adaptive ventilatory control, real-time monitoring, and data-driven respiratory signal analysis.