以輪廓分析與模組度Q揭露台灣前2%醫界科學家之學門與機構構群聚（2025）：量化網絡視覺圖

Revealing Disciplinary and Institutional Clusters among Taiwan's Top 2% Medical Scientists (2025) Using Silhouette Plots and Modularity Q: A Quantitative Network Visualization

美國史丹佛大學最新發布2025年「全球前2%頂尖科學家」名單，受到醫界的矚目。但其常以入榜人數排行，缺乏可解釋與可比較的量化視覺圖。本研究（1）利用輪廓圖（SS）與網絡圖揭示台灣醫界學者的主要學門、機構、與代表學者的成就，並驗證（2）台大前2%頂尖科學家人數排名較泰晤士世界大學的140排名之後（介於140至200間，因前者另含非大學的機構，如醫院與中研院等）。下載台灣醫界入榜名單；（1）以學門與機構共字關係構建網絡；（2）用論文量與被引量篩選主要機構與學者；（3）對代表學者製作個人論文集摘要。輪廓圖輔助網絡圖，另含評估指標：SS、模組度Q*（=Q_w/Q_max，加權Q／最大可達Q）、與絕對優勢係數（AAC）。結果顯示，（1）共316位學者、65個機構、及17個學群；（2）5大學群為癌症、中醫、心臟、藥學、及護理；（3）主要機構為中國醫藥大學、長庚醫院、與台大醫院；（4）被引量最高者為范德威夫．法蘭斯（馬偕，75,494次），論文量最高者為薛博仁（中國醫藥大學，1,018篇）；兩者AAC分別0.58與0.81；集群品質指標分別為SS=0.68、Q*=0.84與SS=0.72、Q*=0.60；台大排名第174位，落在140至200名間。量化的輪廓圖輔助傳統網絡圖呈現「群內一致性」與「群間分離度」，有效突顯台灣醫界前2%學者的學門聚落、核心機構、與代表人物，具備決策與監測的實用價值。

The 2025 Stanford/Ioannidis "Top 2% Scientists" list has drawn wide attention in Taiwan's medical community, yet discussions often rely only on headcounts without interpretable visual analyses. This study applied silhouette plots (SS) and network visualizations to reveal disciplinary structures, institutional clusters, and representative scholars among Taiwan's top 2% medical scientists, and examined why National Taiwan University (NTU) ranks between 140 and 200 globally by headcount-lower than its Times Higher Education ranking (140)-because the 2% list also includes hospitals and research institutes. The Taiwan roster was analyzed through a field-institution co-occurrence network to identify key institutions and scholars based on publication and citation counts. Results showed that 316 scholars from 65 institutions formed 17 clusters across five major fields: cancer, traditional Chinese medicine, cardiology, pharmacology (pharmacy), and nursing. The leading institutions were China Medical University, Chang Gung Memorial Hospital, and National Taiwan University Hospital. The most-cited scholar was Frans Van de Werf (75,494 citations; AAC = 0.58), and the most prolific was Po-Ren Hsueh (1,018 papers; AAC = 0.81). Cluster metrics were SS = 0.68, Q*= 0.84 and SS = 0.72, Q* = 0.60. Integrating silhouette and network visualization effectively captures disciplinary cohesion, institutional influence, and leadership patterns among Taiwan's top 2% scientists.