Evaluating Treatment Response in Patients with Lung Cancer using Electronic Health Records of Multiple Hospitals by Natural Language Processing of Unstructured Data

DOI
  • ARAKI KENJI
    Patient Advocacy Center, University of Miyazaki Hospital
  • MATSUMOTO NOBUHIRO
    Division of Respirology, Rheumatology, Infectious Diseases, and Neurology, Department of Internal Medicine, University of Miyazaki
  • TOGO KANAE
    Health & Value, Pfizer Japan Inc.
  • YONEMOTO NAOHIRO
    Health & Value, Pfizer Japan Inc.
  • OHKI EMIKO
    Oncology Medical Affairs, Pfizer Japan Inc.
  • XU LINGHUA
    Health & Value, Pfizer Japan Inc.
  • HASEGAWA YOSHIYUKI
    Manufacturing IT Innovation Sector, NTT DATA Corporation
  • INOUE HIROFUMI
    Manufacturing IT Innovation Sector, NTT DATA Corporation
  • TAKEMOTO RYOTA
    Manufacturing IT Innovation Sector, NTT DATA Corporation
  • MIYAZAKI TAIGA
    Division of Respirology, Rheumatology, Infectious Diseases, and Neurology, Department of Internal Medicine, University of Miyazaki

Bibliographic Information

Other Title
  • 多施設電子カルテデータベースを用いた肺がん患者における薬物治療効果の評価：非構造化データの自然言語処理

Search this article

Description

<p>　We generated methods for evaluating clinical outcomes including treatment response in oncology using the unstructured data from electronic health records (EHR). This retrospective analysis used medical record database of University of Miyazaki Hospital and the Life Data Initiative EHR database of multiple hospitals in patients with lung cancer who are treated with anticancer therapy. For assessing key terms to describe treatment response (objective response [OR], stable disease [SD] or progressive disease [PD]) adjudicated by evaluators, natural language processing (NLP) rules were created. The most common key terms to describe OR were “reduction/shrink”, “(limited) effect”, “remarkable change”, “improvement”, etc., and “reduction/shrink” showed high sensitivity and specificity. These key terms were also found in the EHR database of multiple hospitals by a systematic review approach. This indicates that assessing response to anticancer therapy in patients with lung cancer using unstructured data of EHRs is feasible.</p>

Journal

Keywords

Details 詳細情報について

Report a problem

Back to top