Particle Swarm Optimization (PSO) Ensembled Neural Networks (ENN) Applied in Pain Evaluation of Post-operation via Patient Controlled Analgesia
| Autor: | Chien-Kuang Wu, 吳建廣 |
|---|---|
| Rok vydání: | 2010 |
| Druh dokumentu: | 學位論文 ; thesis |
| Popis: | 98 Pain is a form of physiological effect that people always hate. The post-operative pain produced by direct impact on people suffering painful and recovery of patients. In the past, medical staff found using anesthetics could reduce pain, but using anesthetics over the standard dose makes patients feel uncomfortable, and even puts them at risk. With the aid new technology, Hospira developed a patient controlled analgesia (PCA) device. PCA performs to be helpful for pain management and the event recording of pain-relief demands and analgesia agent deliveries without administrations by medial stuffs. Recently, the i-pain system has been developed for storing the PCA recordings, which contain the recipes of analgesia agents and the records of PCA events, to benefit evaluations of pain intensity and pain managements. The aim of this study is to introduce the scoring models of pain intensity via the PCA recordings of the i-Pain system. The PCA recordings for postoperative pain managements of gynecological patients treated by the same surgical operations in Wu Ho-Su memorial hospital were used in this investigation. Techniques of artificial intelligent modeling based on neural network were used for building the scoring model of pain intensity. In this study, Particle Swarm Optimization Ensembled Neural Networks (PSO-ENN) for 10 inputs and one output were used for prediction visual analog scale (VAS) of pain intensity in 323 patients receiving intravenous PCA after Caesarean section. Besides, EANN model was calculated for comparison with PSO-ENN prediction results. The prediction performance of PSO-ENN model was satisfactory by root mean square error (RMSE) of training (1.553) and testing (1.821) against EANN model of training (1.6681) and testing (1.99) because visual analog scales were at the range of 0 to 10. The successful results suggest that PSO-ENN model can provide useful reference for the clinical practice in acute pain service. |
| Databáze: | Networked Digital Library of Theses & Dissertations |
| Externí odkaz: |
|