資料探勘技術於嚴重藥物不良反應之探測研究

Abstract

現今，有越來越多吃了食品與藥物管理署認證過的藥物後而得到未預期的嚴重藥物不良反應的案例。嚴重的藥物不良反應狀況有包含死亡、生命危急、需要住院或是延長住院時間、長期或顯著的身體殘障、致畸胎以及其他任何會導致前面五種狀況的疾病。醫院或是病人如果有發現這些嚴重不良反應可以向通報系統通報，而這些藥物就會被重新檢驗。這篇研究主要是希望能夠偵測嚴重藥物不良反應來改善目前的效率以及低估問題。為了資料的完整性，這篇研究選擇使用台灣衛生署健保局的健保資料庫來當我們的資料庫。我們希望能利用關連規則從健保資料庫中找出藥物以及未預期的嚴重藥物不良反應之間的關係。建立的規則會以leverage以及unexlev值來做篩選。在結果上，我們發現有兩種藥物, cisapride以及terfenadine, 所導致的嚴重不良反應可以比通報系統更早被我們偵測到。另外，從丹麥藥品局的報告中，高頻率會導致嚴重藥物不良反應的藥物也在我們篩出來的規則中比較前面的名次。這篇研究所篩選出來的規則可以經由專家驗證後，提早警示食品與藥物管理署來對這些藥物重新做檢驗。

Nowadays, more and more people occurs unexpected serious reaction after taking an FDA-approved drug. Reactions such as death, life-threatening, requires inpatient hospitalization or prolongation of existing hospitalization, persistent or significant disability/incapacity, a congenital anomaly/birth defect, or other situations will be reported to the hospitals and the drugs may be examined again. This study intends to discover the potential unexpected serious ADRs automatically by using the data mining techniques in order to improve the efficiency of detecting and to avoid the under-reporting biases. For the completeness of every patients’ records, we chose the NHIRD as our database. We want to find the strong links between the drugs, which the patients took, and the unexpected serious ADRs, which the patients suffer after taking the drug, by the association rules. Rules would be obtained and chose according to the leverage and unexlev threshold. We found that the serious ADRs of cisapride and terfenadine can be detected earlier than reporting system. The high frequency of drugs that would cause serious ADRs listed by the Danish Medicines Agency's network were found in a high rank. Experts may examine the rules we selected and alarm the FDA for these highlighted relationships.