目的：一、利用效益指標比較婦科癌症病人焦慮與憂鬱症在類神經網路、邏輯斯迴歸及比例風險迴歸模式之準確性。二、利用全域敏感度分析婦科癌症患者焦慮與憂鬱症重要預測因子之權重。方法：資料來源為1996~2010年全民健保資料庫。以年齡≧18歲、ICD-9-CM代碼180、182、183為研究對象，共6,801人，並利用病人特性、醫療機構特性來探討婦科癌症病人焦慮或憂鬱症之重要影響因子。結果：結果顯示類神經網路預測模式優於邏輯斯迴歸（LR）與比例風險迴歸（COX）預測模式。婦科癌症病人焦慮或憂鬱症的準確性，在敏感性部份類神經網路、邏輯斯迴歸（LR）與比例風險迴歸（COX）分別為67.23%、2.90%與52.02%；陰性預測值（NPV）分別為84.04%、66.91%與0.00%；準確性（Accuracy）分別為81.19%、66.36%與17.32%；在AU-ROC曲線分別為88%、53%與26%。結論：研究結果證明類神經網路預測模式是優於邏輯斯迴歸和比例風險迴歸模式。建議醫學研究可運用此模式做為臨床評估及決策參考，找出最重要的預測因子，提升醫療品質。

Objectives: 1. To compare the performance indices between artificial neural networks (ANN), multiple logistic regression (MLR), and Cox proportional hazards (COX) models; and 2. To conduct a global sensitivity analysis to identify the significance of these weighted predictors. Methods: This a retrospective study using the National Health Insurance Research Database from 1996-2010. The study involved 6801 gynecologic cancer patients ＞ 18 years of age, with ICD-9-CM diagnosis codes 180, 182, and 183, and use patient characteristics and hospital characteristics were used to identify the impact of anxiety and depressive disorders. Results: The accuracy of ANN, MLR and COX was 81.19%, 66.36% and 17.32%, respectively. The corresponding area under the receiver operating characteristic curve (AU-ROC) was 88%, 53% and 26%, the sensitivity was 67.23%, 2.90%, and 52.02%, and the NPV was 84.04%, 66.91%, and 0.00%, respectively. The ANN model showed better performance indices for anxiety and depressive disorders than the MLR and COX models. Conclusions: The ANN model was shown to be superior to the MLR and COX models. Therefore, medical research can use this model as a clinical assessment and decision-making reference to identify the most important predictors for improving the quality of medical care.