節律異常之橈動脈脈搏波非線性動態特徵之量化分析

Abstract

本研究旨在針對節律異常之脈搏波訊號，建立具辨識能力與生理意涵之非線性量化指標。傳統分析方法如時域與頻域分析，雖廣泛應用於心血管訊號，卻高度仰賴波形完整性，對於結構紊亂的脈搏波往往難以提供穩定且具解釋力的結果。有鑑於此，本研究聚焦於節律異常且結構混亂的脈搏波，探究其中是否潛藏可量化且具辨識潛力的非線性動態特徵。本研究以非線性分析工具中的遞迴圖、最大李亞普諾夫指數（MLE）與龐加萊映射為核心方法，系統性分析健康、心律不整與臨終三組對象之脈搏波動態特性，特別聚焦於心律不整與臨終患者之異常波形。研究結果顯示，多數遞迴圖參數（RR、DET、ENTR、Lmax）能有效反映系統週期性，而垂直線參數（LAM、TT）未能展現停滯性特徵，反與週期性變化相關，顯示其作為停滯性指標的適用性有限。而MLE與龐加萊映射亦能輔助評估脈搏波系統的混沌程度與波形週期性。綜合所有指標進行機器學習分類，三種多分類模型之分類精確度達82 %，此結果顯示非線性量化指標具一定程度的脈搏波分類能力。這些指標不僅在學術層面上補足傳統分析在異常節律脈搏波處理上的解釋侷限，亦有望在臨床應用中提供具量化依據與生理解釋能力的判別工具，有助於心律不整之診斷與病情追蹤，以及臨終狀態的早期預警，提升臨床決策的精確性與反應效率。

This study aims to develop nonlinear quantitative indices with both discriminative power and physiological relevance for analyzing pulse wave signals characterized by rhythm abnormalities. Traditional time- and frequency-domain analyses are widely used in cardiovascular signal processing but rely heavily on waveform integrity, often leading to unreliable results for structurally disordered pulse waves. To address this limitation, the present study focuses on pulse signals with rhythm abnormalities and disrupted structures, particularly those from patients with arrhythmia and individuals in terminal stages, to investigate whether they contain quantifiable nonlinear dynamic features. Key analytical methods include recurrence plots, the maximum Lyapunov exponent (MLE), and Poincaré maps, applied to data from healthy , arrhythmia patients, and terminally ill patients. Results show that most recurrence-based parameters (RR, DET, ENTR, Lmax) effectively reflect signal periodicity, while vertical line parameters (LAM, TT) do not indicate laminarity as expected and instead show associations with periodic changes, suggesting limited suitability as laminarity markers. MLE and Poincaré maps further assist in characterizing chaotic behavior and rhythm stability. Integrating these indices into machine learning models yielded a classification accuracy of 82%, demonstrating their potential for distinguishing physiological conditions.These findings not only extend the analytical scope beyond conventional methods but also offer clinically meaningful tools for diagnosing arrhythmia, tracking disease progression, and providing early warning of terminal decline, thereby improving the precision and timeliness of clinical decisions.