研究背景：
在這資訊爆炸的時代，醫師必須利用實證醫學的技巧，來獲得最新最好的醫學知識並應用在病人身上。醫學資訊對於實證醫學的研究，過去的文獻大多著重在改善搜尋策略和搜尋引擎的部分。我們提出一個不同於以往文獻報告的新方法，來加速實證醫學的實行。

研究目標：
本研究的目標是發展一個可自動將醫學文獻摘要轉換成實證醫學PICO架構的智慧型系統。我們也評估了是否本系統能夠幫助醫師節省閱讀醫學文獻摘要的時間。

研究方法：
我們利用十個搜尋關鍵字和特殊搜尋策略，從PubMed下載489篇醫學文獻的摘要。然後我們歸納出從摘要中擷取含有PICO資訊的句子之規則，以建立此一系統。我們也測量了醫師閱讀原始摘要和PICO架構的時間，並記錄醫師給予本系統的評分（1-5分）。

研究結果：
本智慧型系統判斷醫學文獻分類的正確率為89%，擷取含有P、I&C、O資訊的句子準確率分別為91.7%、91.6%、87.5%。我們使用獨立T檢定和Mann-Whitney檢定，來比較醫師閱讀原始摘要和PICO架構的時間，結果都顯示閱讀PICO架構的時間比較短，而且有統計學上顯著的差異(P<0.05)。在系統呈現的PICO架構中，醫師能夠讀懂其中的87.1%。醫師給予本系統的平均分數為3.973分。

結論：
本系統成功地將醫學文獻摘要轉換成PICO架構，並能幫助醫師更有效率地閱讀文獻摘要。

Background:
In the times of information explosion, physicians have to use skills of evidence-based medicine (EBM) to acquire the latest and best medical knowledge and apply it to the patients. Most of the previous studies of EBM in the field of medical informatics focused on improving search strategies and search engines. The researcher presented a new method which is different from the previous studies to accelerate the execution of EBM.

Objective:
The aim of this study was to develop an intelligent system to automatically transform medical abstracts into EBM-based PICO frameworks. The researcher also evaluated whether this system could help physicians to save time on reading medical abstracts.

Methods:
The researcher used 10 search keywords with special search strategies to download abstracts of 489 medical citations from PubMed. Then the researcher concluded rules to retrieve sentences including PICO information in the abstracts to establish this system. Time on reading original abstracts and PICO frames was measured, and the scores (1-5) of the system graded by physicians were also recorded.

Results:
The accurate rate of this system to judge classification of medical abstracts was 89%. The precision to retrieve sentences including P, I&C, O information was 97.1%, 91.6% and 87.5%, respectively. The researcher used independent T test and Mann-Whitney test to compare time on reading original abstracts and PICO frames. All results revealed that time on reading PICO frames was significantly shorter (P<0.05) than original abstracts. Physicians could understand 87.1% of the PICO frames presented by this system. The average score of the system graded by physicians was 3.973.

Conclusion:
This system successfully transformed medical abstracts into EBM-based PICO frameworks, and facilitated physicians to read abstracts more efficiently.

[1] MEDLINE/PubMed Resources [Online]. 2008. Available at: http://www.nlm.nih.gov/bsd/index_stats_comp.html Accessed July 1, 2008.
[2] Alper BS, Hand JA, Elliott SG, Kinkade S, Hauan MJ, Onion DK, et al. How much effort is needed to keep up with the literature relevant for primary care? J Med Libr Assoc. 2004;92:429-37.
[3] De Groote SL, Dorsch JL. Measuring Use Patterns of Online Journals and Databases. J Med Libr Assoc. 2003;91:231–41.
[4] 何謂實證醫學, 林口長庚實證醫學中心 [Online]. 2008. Available at: http://www.cgmh.org.tw/intr/intr2/ebmlink/html/define.html Accessed July 1, 2008.
[5] 實證臨床指引平台, 國家衛生研究院 [Online]. 2008. Available at: http://ebpg.nhri.org.tw/Module/Content.aspx?catalog=96 Accessed July 1, 2008.
[6] Critical Appraisal, Centre for Evidenced-Based Medicine (CEBM) [Online]. 2008. Available at: http://www.cebm.net/index.aspx?o=1157 Accessed July 1, 2008.
[7] ACP Journal Club [Online]. 2008. Available at: http://www.acpjc.org/ Accessed July 1, 2008.
[8] Evidence-Based Medicine [Online]. 2008. Available at: http://ebm.bmj.com/ Accessed July 1, 2008.
[9] The Cochrane Collaboration [Online]. 2008. Available at: http://www.cochrane.org/ Accessed July 1, 2008.
[10] Haynes RB, Wilczynski N, McKibbon KA, Walker CJ, Sinclair JC. Developing optimal search strategies for detecting clinically sound studies in MEDLINE. J Am Med Inform Assoc. 1994;1:447-58.
[11] Summary of Enhancements for Clinical Queries for MEDLINE for Studies [Online]. 2008. Available at: http://www.nlm.nih.gov/pubs/techbull/jf04/cq_info.html Accessed July 1, 2008.
[12] Clinical Queries Filter Table [Online]. 2008. Available at: http://www.ncbi.nlm.nih.gov/entrez/query/static/clinicaltable.html Accessed July 1, 2008.
[13] Schardt C, Adams MB, Owens T, Keitz S, Fontelo P. Utilization of the PICO framework to improve searching PubMed for clinical questions. BMC Med Inform Decis Mak. 2007;7:16. DOI 10.1186/1472-6947-7-16
[14] Aphinyanaphongs Y, Tsamardinos I, Statnikov A, Hardin D, Aliferis CF. Text categorization models for high-quality article retrieval in internal medicine. J Am Med Inform Assoc. 2005;12:207-16.
[15] Fiszman M, Rindflesch TC, Kilicoglu H. Abstraction Summarization for Managing the Biomedical Research Literature. Proceedings of the HLT-NAACL Workshop on Computational Lexical Semantics. East Stroudsburg, PA: Association for Computational Linguistics, 2004, pp. 76–83.
[16] Demner-Fushman D, Lin J. Answering clinical questions with knowledge-based and statistical techniques. Comput Linguist. 2007;33:63–104.
[17] Sneiderman CA, Demner-Fushman D, Fiszman M, Ide NC, Rindflesch TC. Knowledge-based methods to help clinicians find answers in MEDLINE. J Am Med Inform Assoc. 2007;14:772-80.
[18] 醫藥衛生研究資訊網 [Online]. 2008. Available at: http://www.hint.org.tw/dbguide/medline.htm Accessed July 1, 2008.
[19] MEDLINE Fact Sheet [Online]. 2008. Available at: http://www.nlm.nih.gov/pubs/factsheets/medline.html Accessed July 1, 2008.
[20] PubMed Overview [Online]. 2008. Available at: http://www.ncbi.nlm.nih.gov/entrez/query/static/overview.html Accessed July 1, 2008.
[21] PubMed Help [Online]. 2008. Available at: http://www.ncbi.nlm.nih.gov/books/bv.fcgi?rid=helppubmed.chapter.pubmedhelp Accessed July 1, 2008.
[22] NLM Standard Publisher Data Format [Online]. 2008. Available at: http://www.ncbi.nlm.nih.gov/entrez/query/static/spec.html Accessed July 1, 2008.
[23] Unified Medical Language System [Online]. 2008. Available at: http://www.nlm.nih.gov/research/umls/ Accessed July 1, 2008.
[24] 統一醫學語言系統簡介 [Online]. 2008. Available at: http://www.lib.ntu.edu.tw/pub/mk/mk27/mk27_1.html Accessed July 1, 2008.
[25] Aronson AR. Effective mapping of biomedical text to the UMLS Metathesaurus: the MetaMap program. Proc AMIA Symp. 2001;17-21.
[26] MetaMap Transfer (MMTx) [Online]. 2008. Available at: http://mmtx.nlm.nih.gov/ Accessed July 1, 2008.
[27] Pratt W, Yetisgen-Yildiz M. A study of biomedical concept identification: MetaMap vs. people. AMIA Annu Symp Proc. 2003;529-33.
[28] Armstrong EC. The well-built clinical question: the key to finding the best evidence efficiently. WMJ. 1999;98:25-8.
[29] American Family Physician [Online]. 2008. Available at: http://www.aafp.org/online/en/home/publications/journals/afp.html Accessed July 1, 2008.
[30] Family Physicians Inquiries Network (FPIN) [Online]. 2008. Available at: http://www.fpin.org/ Accessed July 1, 2008.
[31] PrimeAnswers Project [Online]. 2008. Available at: http://healthlinks.washington.edu/primeanswers/project/ Accessed July 1, 2008.
[32] PrimeAnswers [Online]. 2008. Available at: http://primeanswers.org/primeanswers/ Accessed July 1, 2008.
[33] PrimeAnswers (documents from FPIN Clinical Inquiries) [Online]. 2008. Available at: http://primeanswers.org/primeanswers/search/parentID/results.cfm?parentResourceID=294 Accessed July 1, 2008.
[34] 行政院衛生署衛生統計資訊網 [Online]. 2008. Available at: http://www.doh.gov.tw/statistic/index.htm Accessed July 1, 2008.
[35] Ad Hoc Working Group for Critical Appraisal of the Medical Literature. A proposal for more informative abstracts of clinical articles. Ad Hoc Working Group for Critical Appraisal of the Medical Literature. Ann Intern Med. Apr;106(4):598-604, 1987
[36] SPSS [Online]. 2008. Available at: http://www.spss.com/ Accessed July 1, 2008.
[37] Armitage P, Berry G, Matthews JNS. Statistical Methods in Medical Research. Hoboken, NJ: Blackwell Science; 2002.
[38] Parsons HM. What Happened at Hawthorne? New evidence suggests the Hawthorne effect resulted from operant reinforcement contingencies. Science. 1974;183:922-32.