The methodology of facilitating data analysis in medical informatics -information extraction from free-text data and structural data collection through the structure report interface

臨床上的醫療資料含有大量的病患資料，如手術記錄、出院病摘以及各種檢查的報告，而這些資料中，可能含有許多寶貴的資訊。如果能進一步分析這些資料，則可以有機會在這大量的醫療資料中得到有用的知識。 一般來說，臨床資料可以被分為兩種類型:經過結構化的資料以及沒有經過結構化的資料.對結構化的資料來說，這些資料是可以直接被電腦來分析，但是非結構化的資料在被分析之前則需要先被經過額外的處理以及抽取。所以我們要解決的最主要的問題就是如何將資料結構化。針對這個議題，我們採取兩種方法來達到資料結構化的目的。 1.針對未來的報告資料: 設計了結構化報告輸入介面讓使用者可以使用此介面去收集肝癌的臨床資料。 2.針對過往的歷史資料: 在這篇論文中我們開發了一個資料萃取輔助系統，其最主要的功能就是可以從結構鬆散(Free-text)的醫學報告中去抽取出有用的資訊。 藉由結合這兩種功能，醫師可以結構化肝癌方面的臨床資料以利於日後相關主題的研究。Clinical data consist of abundant patients data such as operation notes, diagnosis, and various examination reports. These clinical data may contain a rich source of valuable information. If we can further analyze the clinical data, then the useful knowledge may be obtained from the huge amount of medical data. In general, the clinical data can be divided into two types: the structuralized data and the non-structuralized data. For the structuralized data, the data can be analyzed by computer directly, however, the non-structuralized data need to be additionally processed and extracted before the further studying. Therefore the major problem we want to solve is how to structuralize the clinical data. We adopt two ways to fulfill the target for structuralizing data. 1. Reports in the future: A Structure Report Interface is designed for collecting liver cancer clinical data by this interface. 2. Clinical documents in the past: The Data Extraction Assistant System is designed for extracting useful information from the medical free-text report. Through the combination of these two functions, physicians can structuralize liver cancer clinical data and we hope the system can facilitate the study relevant to liver cance