English  |  正體中文  |  简体中文  |  全文筆數/總筆數 : 45422/58598 (78%)
造訪人次 : 2521798      線上人數 : 176
RC Version 7.0 © Powered By DSPACE, MIT. Enhanced by NTU Library IR team.
搜尋範圍 查詢小技巧:
  • 您可在西文檢索詞彙前後加上"雙引號",以獲取較精準的檢索結果
  • 若欲以作者姓名搜尋,建議至進階搜尋限定作者欄位,可獲得較完整資料
  • 進階搜尋
    請使用永久網址來引用或連結此文件: http://libir.tmu.edu.tw/handle/987654321/58421


    題名: ICD-10疾病編碼輔助系統:基於多標籤分類的處方藥物資訊
    Computer-assisted ICD-10 coding system: Based on prescribed medications information by multi-label classification
    作者: 郭偵丈
    Kuo, Chen-Cheng
    貢獻者: 醫學資訊研究所
    林明錦
    關鍵詞: 疾病診斷碼;多標籤文本分類
    Diagnosis codes;Multi-label text classification
    日期: 2019-01-02
    上傳時間: 2020-01-14 12:07:09 (UTC+8)
    摘要: 根據病人情況編寫相對應的疾病診斷碼於其電子病歷中,是臨床工作中常規的作業程序。然而,有鑒於國際疾病診斷分類碼規則具有其複雜性,臨床人員常需花費時間於適當診斷碼的挑選。為減少時間與人力的耗費,加速與提升編碼過程與完整性,我們希望能建立分類模型藉由醫生所開立之藥物資訊即時地去建議相關對應之疾病診斷碼。

    隨著深度學習在眾多領域中的突破性發展,文本分類相關研究亦受到此潮流所影響,目前已有越來越多的研究投入深度學習模型之開發。在本研究中,我們參考Yoon kim 在2014年研究中所提出的方法,先利用Word2vec對語料庫(corpus)中的文字資訊生成向量,再使用卷積神經網路進行分類。考量到不同人員間可能存在的編碼變異性,我們將資料集切割成單位醫生之子集分別進行模型的訓練與測試。

    結果顯示,本研究中所訓練之模型在Recall的表現上都相當良好,近七成的醫生模型分類之Recall值可達到0.8以上,意味著該模型在疾病診斷碼分類建議上可涵蓋八成以上於原始病歷中所紀錄之診斷碼。基於所使用的資料在疾病編碼上可能存在Under-coding之現象,而間接使得Precision表現受到低估,我們對分類結果中的分類錯誤之偽陽性標籤進行分析,透過此分析發現數個偽陽性實則應為真陽性分類之案例,顯示此套方法所建立之模型對於疾病診斷碼編寫之完整性是有所提升的。
    Assigning diagnosis codes to electronic medical records (EMRs) is an essential task in healthcare facilities. However, with the complexity of ICD-10-CM, it takes time for physicians or coders to select appropriate codes. To reduce the consumption of time and human labor on assigning diagnosis codes, we try to construct a computer-assisted coding system to facilitate the process of assigning diagnosis codes.

    With the successful development of deep learning in many domains, research of text classification is also affected by this trend. In this research, we referred by the approach proposed by Yoon Kim in 2014 research. We got the vector representations of words in our corpus from Word2Vec and used a simple convolutional neural network for classification. Due to the issue of coding variance, we subset our data into one doctor to train and test each model separately.

    The models showed their ability on automatically suggesting the most related diagnosis codes based on the information of prescribed medications. From the result, we found that recall performed well in most doctors testing, there are about 70% classifications of the doctor models whose recall values can reach 0.8 or more meaning that our models can cover 80% or more diagnosis codes recorded in original medical record. Due to the issue of under-coding of diagnosis codes ,we concerned that the performance of precision may be underestimated, so we did the error analysis about false positives. We found the models is helpful to improve the completeness of diagnosis codes through the analysis of false positive values.
    描述: 碩士
    指導教授:林明錦
    委員:徐建業
    委員:蘇家玉
    資料類型: thesis
    顯示於類別:[醫學資訊研究所] 博碩士論文

    文件中的檔案:

    檔案 描述 大小格式瀏覽次數
    index.html0KbHTML92檢視/開啟


    在TMUIR中所有的資料項目都受到原著作權保護.

    TAIR相關文章

    著作權聲明 Copyright Notice
    • 本平台之數位內容為臺北醫學大學所收錄之機構典藏,包含體系內各式學術著作及學術產出。秉持開放取用的精神,提供使用者進行資料檢索、下載與取用,惟仍請適度、合理地於合法範圍內使用本平台之內容,以尊重著作權人之權益。商業上之利用,請先取得著作權人之授權。

      The digital content on this platform is part of the Taipei Medical University Institutional Repository, featuring various academic works and outputs from the institution. It offers free access to academic research and public education for non-commercial use. Please use the content appropriately and within legal boundaries to respect copyright owners' rights. For commercial use, please obtain prior authorization from the copyright owner.

    • 瀏覽或使用本平台,視同使用者已完全接受並瞭解聲明中所有規範、中華民國相關法規、一切國際網路規定及使用慣例,並不得為任何不法目的使用TMUIR。

      By utilising the platform, users are deemed to have fully accepted and understood all the regulations set out in the statement, relevant laws of the Republic of China, all international internet regulations, and usage conventions. Furthermore, users must not use TMUIR for any illegal purposes.

    • 本平台盡力防止侵害著作權人之權益。若發現本平台之數位內容有侵害著作權人權益情事者,煩請權利人通知本平台維護人員([email protected]),將立即採取移除該數位著作等補救措施。

      TMUIR is made to protect the interests of copyright owners. If you believe that any material on the website infringes copyright, please contact our staff([email protected]). We will remove the work from the repository.

    Back to Top
    DSpace Software Copyright © 2002-2004  MIT &  Hewlett-Packard  /   Enhanced by   NTU Library IR team Copyright ©   - 回饋