SHIP: A Shapelet-based Approach for Interpretable Patient-Ventilator Asynchrony Detection

TL;DR

PVA is a common, clinically important problem in mechanical ventilation, often hampered by data imbalance and opaque models. SHIP introduces a shapelet-based pipeline that includes Offline Shapelet Discovery, shapelet-based augmentation, PSD-based shapelet transform, and LogSig statistics to achieve accurate, interpretable PVA detection. Experimental results on a clinically collected dataset show SHIP achieves state-of-the-art per-class F1 and overall accuracy, and can operate effectively with only two channels (Pmask and Flow). The method provides interpretable insights by linking decisions to distances to recognizable shapelets, supporting clinical decision-making and potential real-time monitoring.

Abstract

Patient-ventilator asynchrony (PVA) is a common and critical issue during mechanical ventilation, affecting up to 85% of patients. PVA can result in clinical complications such as discomfort, sleep disruption, and potentially more severe conditions like ventilator-induced lung injury and diaphragm dysfunction. Traditional PVA management, which relies on manual adjustments by healthcare providers, is often inadequate due to delays and errors. While various computational methods, including rule-based, statistical, and deep learning approaches, have been developed to detect PVA events, they face challenges related to dataset imbalances and lack of interpretability. In this work, we propose a shapelet-based approach SHIP for PVA detection, utilizing shapelets - discriminative subsequences in time-series data - to enhance detection accuracy and interpretability. Our method addresses dataset imbalances through shapelet-based data augmentation and constructs a shapelet pool to transform the dataset for more effective classification. The combined shapelet and statistical features are then used in a classifier to identify PVA events. Experimental results on medical datasets show that SHIP significantly improves PVA detection while providing interpretable insights into model decisions.

Figures (2)

• Figure 1: The general architecture of our proposed method for detecting PVA events (Autocycling (AC), Double Triggering (DT), and Ineffective Efforts (IE)) and Non-PVA (NP).
• Figure 2: The interpretability of shapelets for PVA detection, including four types of events (NP, AC, DT, IE).