利用資料探勘技術預測術後噁心嘔吐

Abstract

術後噁心嘔吐（Postoperative nausea and vomiting，PONV）為手術後常見的副作用之一，而其發生受到多重因素的影響，包含了病患生理狀況、手術因素及麻醉的使用等，而病人疼痛自控裝置(Patient-Controlled Analgesia ,PCA) 是可由病患自行操控的術後麻醉方式，已逐漸普遍使用於各大醫院中。本研究使用資料探勘技術，參考各種因素以及病人疼痛自控裝置之使用行為，結合監督與非監督學習演算法來預測病患術後噁心嘔吐之副作用，實驗結果顯示本研究提出之方法確實可提升預測效果。此預測可提供醫護人員作為參考，即時給予病患副作用預防性藥物，免除病患術後復原中不必要的痛苦並提升術後照護品質。

Postoperative nausea and vomiting (PONV) occurs as the common side effect of anesthesia. Risk factors include Patient-Related Factors, Anesthesia-Related Factors and Surgery-Related Factors. Patient Controlled Analgesia (PCA) is a method allowing patients to self-administer analgesics commonly used in hospitals.This study aimed to predict postoperative nausea and vomiting, and considers PCA demand behavior. The results of the experiment indicated that our method can achieve better performance by using the proposed method. The prediction will be available to the healthcare professional attribute reference, and improved both medication safety and the quality of postoperative care.