以文字探勘技術應用於疾病分類之輔助系統-以出入院病歷摘要為例
Applies by Text Mining to the Support Systems for Coding
ICD-9-CM —A study of Admission Note and Discharge Summary
中文摘要
國際疾病分類系統是全球公共衛生界用以描述疾病、分析病歷及訂定衛生政策時 經常使用的溝通工具。而疾病分類的工作是由醫院病歷單位的疾病分類人員閱讀 病歷，並評估疾病資料之適當性，以確認主要診斷、次要診斷、處置及併發症，
並依照 ICD-9-CM 手冊上準則轉換成適當的數字代碼，而早期記載的出院病歷摘 要文件需靠疾病分類人員解讀進行編碼，然而經由疾病分類人員解讀病歷不但耗 時也容易造成編碼錯誤或遺漏。因此，本研究動機是希望結合文字探勘技術與國 際疾病分類編碼規則，試圖從出入院病歷摘要文件資料中找出描述疾病與編碼所 隱藏的規則，並將此規則應用於疾病分類系統中，並探討此輔助系統與疾病分類 人員編的疾病碼差異性及改善自動文件分類的準確性，藉以協助疾病分類人員能 提升編碼工作效率及節省編碼時間。
英文摘要
The global public health often uses International Classification of Disease to analysis the disease and decides health policy from medical record. The coding medical record is worked by the hospital management medical record unit''s coders who confirm the main diagnosis, the secondary diagnosis, procedure and complication, and refer to in the ICD-9-CM handbook to transform the suitable numeric code. However human coder is also easy to come into wrong code or omitting code. We hope using text mining for mining the transcriptions of dictated medical record. The support system compares the variability among experienced coders. We expect that the support system is as accurate as human coders and offers the potential for increased coding consistency and
productivity.
