探究生命末期病患與皮膚壓瘡之相關性分析

Exploring the Correlation Analysis of End-of-life Patients and Skin Pressure Ulcer

近年來隨著醫療科技的進步，它可以預測死亡是否會到來，但因病程的變化仍存有許多的變異，預測的正確率不易達到可接受的水準。皮膚為人體最大面積的器官，而病人在生命末期時，皮膚會發生衰竭。臨床上常見到生命末期病患於短期中發生壓瘡且其過程進展很快，儘管臨床照顧人員已經建立相關壓瘡防治的管控模式與預防措施，仍無法避免皮膚破裂的發生。本研究以邏輯斯迴歸和決策樹演算法來推估壓瘡發生的可能性，以幫助臨床照顧人員瞭解生命末期病患壓瘡發生的可能原因與意義，並事先對照顧者做好壓瘡防治的衛教。本研究以邏輯斯迴歸和決策樹演算法來探究生命末期病患與皮膚壓瘡之相關性分析，樣本收集期間為2007年1月至2012年11月止共1333筆資料。男生病患為731人、女性病患為602人。有壓瘡病患為664人、無壓瘡病患為669人。透過配合不同參數之設定與交叉驗證，結果求得邏輯斯迴歸最佳預測準確度為73.82%，而決策樹求得最佳預測的準確度為76.14%。與邏輯斯迴歸相比較，決策樹演算法之正確率較高且可產生規則以供應用，因此決策樹演算法比邏輯斯迴歸更適用於探究生命末期病患與皮膚壓瘡之相關性分析。

With the advancement of medical technology nowadays it's now possible to predict the inevitable death. Because there are some variants in the course of disease, the prediction accuracy rate is not acceptable. The skin is the largest organ in our body and skin failure occurs in the dying patients. The pressure ulcer in the dying patients occurs in the short term and the process progressed very quickly. Although the clinical care staff had established some pressure ulcers control methods and preventive measures, it still unable to avoid the occurrence of skin breakdown. This study applied logistic regression and decision tree algorithm to predict the possibility of pressure ulcers, which helps clinical care staff understand the possible reason and meaning of pressure ulcers for dying patients and renders advance care for clinical care staff. In this study, logistic regression and decision tree algorithm are used to correlate the dying patients with pressure ulcers. Skin correlation analysis samples were collected during the period from January 2007 to November 2012 for a total of 1333 documented patients. Among them 731 were male, 602 were female. And 664 patients were with pressure ulcers, 669 patients were without pressure ulcers. Through with different parameters settings and cross validation, the prediction accuracy of the logistic regression is 73.82%, while that of the decision tree is 76.14%. Compared with logistic regression, the decision tree algorithm has higher classification accuracy rate and provides rules for further use, so the decision tree algorithm is more suitable than logistic regression to explore the correlation analysis of end-of-life patients and skin pressure ulcer.