以多項性生理訊號早期預測急性中風病患預後

Abstract

腦中風是世界上造成死亡或失能的主要原因，耗費的醫療照護成本相當巨大。若能盡早預測腦中風病患的復原結果，能夠幫助診斷、及時提供有效的治療。有很多研究試著要找出能夠預測腦中風預後的因子，但目前尚未有明確的定論。 腦中風如果影響到腦幹的部分通常預後較差，因為腦幹負責調節許多重要的機能，包括控制心跳和血壓的自律神經系統。因此，心跳和血壓理應能夠反應中風後腦部的狀況。目前已經有一些研究是從心電圖和血壓訊號中找腦中風預後的預測因子，大部份都採用線性的分析方法，如二十四小時心跳的平均值和標準差。但由於心跳和血壓的調節包含非線性特徵，線性模型可能不足以描述復雜的人體機能。 本論文的目標是要用生理訊號預測急性中風病患的預後。我們提出一套完整的流程，要用心電圖和血壓訊號預測中風病患的三十天後的預後結果，並採用多尺度熵（Multiscale Entropy, MSE）分析方法，觀察信號複雜度和中風預後的關聯。分析結果顯示，預後較好的中風病患，信號複雜度明顯地高出預後較差的中風病患。而且我們只需使用兩個小時的心電圖和血壓訊號，即能找出此結果，這在臨床使用上是較為實際且有效率的。本文的研究結果顯示，用多尺度熵分析心電圖和血壓訊號，可以早期預測急性中風病患的預後。

Stroke is the leading cause of death and disability worldwide. This requires significant resources in health-care costs. Early prediction of stroke outcome can help in providing effective treatment. A lot of studies have analyzed associated predictors and stroke outcome, but there’s no clear conclusion. A stroke affecting the brainstem usually results in a poor outcome, since the brainstem controls several important functions, including the autonomic nervous system, which regulates heart rate (HR) and blood pressure (BP). Therefore, it is believed that HR and BP will reflect brain’s condition after stroke. Existing works tried to find predictors of stroke outcome from electrocardiogram (EKG) and BP signals; most studies used linear models. However, modulations of the HR and BP are thought to be predominantly nonlinear. Linear models might be insufficient to describe these complex dynamics adequately The goal of the thesis is to predict functional outcome of patients with acute stroke based on biophysiological signals. A complete procedure for 30-day outcome prediction based on EKG and BP is proposed. The multiscale entropy (MSE) method is used to analyze the relation between signal complexity and outcome after acute stroke. The analysis results show that signal complexity of stroke patients with good outcome is significantly higher than that of patients with poor outcome. These results can be found in 2-hour recordings of EKG and BP, which is effective and practical in clinical use. This work suggests that MSE analysis of EKG and BP can be early outcome predictors in acute stroke patients.