Machine learning for inferring new drug indications based on clinical information약물 재창출을 위한 임상정보기반 기계학습

Cited 0 time in webofscience Cited 0 time in scopus
  • Hit : 691
  • Download : 0
In last decade, investment in drug research and development(R&D) has steadily increased. However, the number of new drug entities which have been approved by US FDA(Food and Drug Administration) is not proportional to that, because 90% of new drug candidates have been failed in safety test. In this situation, many researchers are taking interests in repurposing or repositioning strategies that is aimed towards finding new indications for existing drugs. Approaches for identifying new drug indications are divided into two major categories which consist of target-based and effect-based approach. However, two approaches contain limitations such as translational and partial problems. In this research, therefore, we present a novel computational framework that consists of three steps to overcome the limitations and suggest reasonable drug candidates for disease. First, disease-specific pathological traits are extracted from large-scale clinical infor-mation(PHRs). Second, PubMed abstracts are used to make drug profiles for disease-specific pathological traits. Finally, we suggest new drug candidates for disease by scoring drug profiles. In order to show how our method works in action, we selected asthma as a case study and inferred 1,225 drug candidates for asthma. We evaluated our inferred results by comparison with previous method(CoPub Discovery Method) on test sets which consist of literature based and CTD set. In comparison with two test set, our method outperforms AUC, Precision, Recall of other method. From this assessment, we may say that disease-specific pathological traits can be considered as features to infer new indications of existing drugs. Therefore, we expect that our method, which is based on clinical information can help infer new and novel candidate for drug repositioning.
Advisors
Lee, Do-Heonresearcher이도헌
Description
한국과학기술원 : 로봇공학학제전공,
Publisher
한국과학기술원
Issue Date
2012
Identifier
509177/325007  / 020104412
Language
eng
Description

학위논문(석사) - 한국과학기술원 : 로봇공학학제전공, 2012.8, [ 6, 48 p. ]

Keywords

Drug repositioning; Pathological trait; Clinical information; literature mining; 신약재창출; 병리학요소; 임상정보; 회기분석; 문헌 마이닝; logistic regression

URI
http://hdl.handle.net/10203/182400
Link
http://library.kaist.ac.kr/search/detail/view.do?bibCtrlNo=509177&flag=dissertation
Appears in Collection
RE-Theses_Master(석사논문)
Files in This Item
There are no files associated with this item.

qr_code

  • mendeley

    citeulike


rss_1.0 rss_2.0 atom_1.0