我們提出使用“meta-search”搜尋引擎在線上即時對醫學文獻做叢集分析來進行實證醫學的新方法。此法對「根據證據的整形外科」理想特別有利。除了把雜亂的搜尋結果整合成有意義而單純的文件群,協助日常的文獻檢索外,我們強調此法可根據資訊技術選擇其他相關的關鍵字詞,並呈現一個主題的「知識構造」,使研究者注意以前未注意到的新課題。 以「乳房重建」的主題搜尋近兩年的PubMed文獻得到496篇文章。線上的叢集分析把這些檢索結果歸類成三十幾個小分群,其中九個較大的分群包括「皮膚保留手術」,「組織擴張器」,「DIEP皮瓣」,「乳頭重建」,「擴背肌」,「穿透支皮瓣」,及「TRAM皮瓣」等常見的次主題。我們繼續探索部分次主題之知識結構,並描述「中國婦女」這群文章的新發現。
We present the application of ”meta-search engine” in evidence-based medicine (EBM) by online cluster analysis of PubMed literature. The methodology would be especially beneficial to ”evidence-based plastic surgery”. In addition to the initial aim of splitting the diverse documents into organized relevant and homogeneous document groups (clusters), knowledge discovery of new or neglected sub-topics and selection of key phrases based on information technology are stressed. PubMed query on ”breast reconstruction” in the recent 2 years yielded 496 articles, which was processed by the online tool. Nine larger clusters comprise of about 70% of the literature, which is ”skin-sparing”, ”tissue expander”, ”DIEP flap”, ”nipple”, ”latissimus dorsi”, ”perforator flap”, ”TRAM flap”, and ”internal mammary vessels”. The details of the knowledge structure were further explored, and a new finding of the ”Chinese women” cluster was described.