乳癌患者化療前後虛擬實境放鬆療效分析：可 解釋式人工智慧之洞見;Analyzing Pre- and Post Chemotherapy VR Relaxation Efficacy in Breast Cancer Patients: Insights from Explainable AI

NCUIR > College of Electrical Engineering & Computer Science > Graduate Institute of Computer Science and Information Engineering > Electronic Thesis & Dissertation > Item 987654321/98538

Please use this identifier to cite or link to this item: https://ir.lib.ncu.edu.tw/handle/987654321/98538

Title:	乳癌患者化療前後虛擬實境放鬆療效分析：可解釋式人工智慧之洞見;Analyzing Pre- and Post Chemotherapy VR Relaxation Efficacy in Breast Cancer Patients: Insights from Explainable AI
Authors:	李洋昊;Lee, Yang-Hao
Contributors:	資訊工程學系
Keywords:	虛擬實境;機器學習;心律變異;皮膚電反應;呼吸變異度;腦電圖;Virtual reality (VR);machine learning (ML);heart rate variability (HRV);galvanic skin response (GSR);respiratory rate variability (RES)
Date:	2025-08-07
Issue Date:	2025-10-17 12:53:59 (UTC+8)
Publisher:	國立中央大學
Abstract:	本乳癌化療經常引發心理與生理上的壓力，進而降低病患的身心福祉。本研究比較在化療前（BC）與化療後（AC）實施虛擬實境（VR）放鬆訓練的成效。研究過程收集了客觀量化放鬆程度的生理訊號，包括腦電圖（EEG）、心率變異度（HRV）、呼吸變異度（RES）及皮膚電反應（GSR）。統計結果顯示，雖然 BC 與 AC 兩組皆呈現放鬆指標的改善， AC 組的變化更為顯著且全面，特別是在自律神經調節相關參數（HRV）、呼吸模式以及部分 EEG 頻段。相關性分析進一步指出，AC 組這些生理上的進步與更大幅度的主觀壓力舒緩相吻合，支持化療後病患對 VR 介入更具接受度的假設。此外，機器學習方法（結合 SHAP 值解釋）凸顯了若干關鍵特徵——如枕葉 α 節律及 HRV 低頻比——在區分介入前後狀態時的影響力。結果顯示，VR 放鬆訓練不僅能緩解心理壓力，也有助於減輕化療相關的身體不適，擴大了其在臨床上超越傳統焦慮管理的應用潛力。未來研究宜納入更大樣本並採縱向設計，以釐清驅動這些效應的機制，深入了解為何病患在化療後從 VR 放鬆中獲益最大，以及多次訓練方案如何最佳化長期效果。;Breast cancer chemotherapy often induces both psychological and physiological stress, leading to decreased patient well-being. This study compares the efficacy of virtual reality (VR)–based relaxation training administered before (BC) versus after (AC) chemotherapy for breast cancer patients. Physiological signals—including electroencephalogram (EEG), heart rate variability (HRV), respiratory variability (RES), and galvanic skin response (GSR)—were collected to objectively quantify re- laxation. Statistical analyses revealed that while both BC and AC groups exhibited improved indicators of relaxation, the AC group demonstrated more pronounced and comprehensive changes, particularly in parameters related to autonomic regulation (HRV), respiratory patterns, and certain EEG frequency bands. Correlation analyses suggested that these physiological improvements in the AC group coincide with greater subjective stress relief, supporting the notion that post-chemotherapy patients may be more receptive to VR interventions. Furthermore, machine learning approaches, augmented by SHAP values, highlighted key features—such as alpha rhythm in occipital regions and HRV low-frequency ratios—that were influential in distinguishing pre- and post-intervention states. These insights indicate that VR relaxation may not only alleviate psychological stress but also help address chemotherapy-related physical discomfort, thereby extending its clinical utility beyond conventional anxiety management. Future work should involve larger samples and longitudinal designs to clarify the underlying mechanisms driving these effects, offering deeper insight into why patients may benefit most from VR relaxation after chemotherapy and how multi- session protocols could optimize long-term outcomes.
Appears in Collections:	[Graduate Institute of Computer Science and Information Engineering] Electronic Thesis & Dissertation

Files in This Item:

File	Description	Size	Format
index.html		0Kb	HTML	26	View/Open

社群 sharing

Loading...