連結個案資料分群與即時文獻搜尋之醫療決策輔助系統

Abstract

面對一個個案，常常有需要參考過去的處理經驗，以及文獻上面的相關資料，雖然可以查但是對於分散在醫療資訊系統以及網際網路上龐大的資料並不方便。可以讓醫師簡單快速的看到先前其他類似個案的診斷治療方式，醫師就能更有信心的評估治療的成效，減少發生失誤的機會，如果還可以針對病患的情形快速的找到相關的文獻證據，以實證醫學的角度(Evidence Based Medicine)連結到與個案最接近的文獻，減少閱讀無關的資料，對於醫師來說是很有幫助的。 我們藉由設計可即時找到過去相同或類似的個案了解其治療處理方式以及將個案最相關的文獻整理好並且列出文章概念與個案的相關性排序，提供醫師實證醫學的決策參考。另外，由於網路速度有限，所以為了改善使用者的回應時間，我們還實做了以代理伺服器快取為基礎的快取機制提升系統效能，最後我們以專家建立標準答案後進行系統驗證，並且測試快取系統之效能。

When physician faced to a case, they often need to refer the past experience and related literature. But it is inconvenient to check these huge Internet information and got similar cases from HIS .If physician can be allowed easily and quickly to see the previous similar cases`s diagnosis and treatment course, they will be able to have more confidence in the treatment and reduce the chance of mistakes. In the point of view of evidence-based medicine, also if we can find the evidence related to the cases is helpful for the physician. We design a system which can find the same or similar cases from database for reference about clinical course and treatment .Also it can display the related literature and be sorted by the relevant to particular cases. These results can help physician to make decision. In addition, because the performance is limited, we also implement a proxy-based cache mechanism to improve user`s response time. We also compare the precision/recall of our system analysis result with the expert`s answer to evaluate the accuracy.