[Background] In the last two decades, the global pharmaceutical industry is facing research & development (R&D) puzzles. The US is now the world's largest pharmaceutical market, while China is the largest emerging market. [Aim] This article aims to compare the growth of listed pharmaceutical companies in China and the US from 2003 to 2018 using a machine learning algorithm. [Methods] According to the R&D capability (e.g., R&D investment ratio) and profitability (e.g., return on equity, ROE), 270 US-listed and China-listed pharmaceutical companies were divided into four levels: Level I (high R&D, high profitability), Level II (high R&D, low profitability), Level III (low R&D, high profitability), Level IV (low R&D, low profitability). These four levels are used to evaluate the growth of pharmaceutical firms. LightGBM algorithm is used to analyze the importance of factors that affect pharmaceutical companies' performance. [Results] LightGBM algorithm was used to build the model with an accuracy of 0.80 for US companies and 0.64 for Chinese companies. The feature importance shows the significance of financial indicators for Chinese and US companies, such as the net profit growth rate. On the other hand, the number of new drugs predominates in the R&D indicators. This is the common ground between Chinese and US pharmaceutical firms, while the growth process of the two is different, From the 1990s to 2010, Some US pharmaceutical companies have grown from Level I to Level I after the mergers and acquisitions (M&A) boom, such as Biogen, Celgene. In the recent decade, biomedical innovation in the US pharmaceutical industry changed to various sources. Chinese pharmaceutical industry grew healthy and rapidly in the past five years. The drug policy reform in China started in 2015, and then China joined the International Conference on Harmonization of Technical Requirements for Registration of Pharmaceuticals for Human Use (ICH) in 2017. Several pharmaceutical companies (e.g., Hengrui, Salubris) entered Level I, and the R&D trend shifted from generic to innovative drugs. [Conclusion] The US will continue to dominate the global pharmaceutical industry. Chinese pharmaceutical industry should seize the opportunity of the rise of machine learning and actively engage in international cooperation to narrow down the gap with the US. Keywords: machine learning, China-US comparison, pharmaceutical industry, growth analysis

背景：在過去的二十年中，全球製藥行業正面臨研發(R&D)難题。美國現在是全球最大的醫藥市場，而中國是最大的新興市場。 目的：本文旨在利用機器學習算法比較 2003 年至 2018 年中美上市製藥公司的成長情況。 方法:根據研發能力(研發投入率)和盈利能力(股本回報率)，将270家美國上市和中國上市製藥公司分為四個等級:I級(高研發能力、高盈利能力)、Ⅱ級(高研發能力、低盈利能力)、Ⅲ級(低研發能力、高盈利能力)、IV級(低研能力、低盈利能力)。這四個級別用於評估製藥公司的成長性。LightGBM 算法用於分析影製藥公司業績的因素的重要性。 結果：採用 LightGBM 算法建立模型，美國公司精度為0.80，中國公司精度為0.64。特徵重要性體現了淨利潤增長率等財務指標對中美企業的重要性。另一方面，新藥數量在研髮指標中占主導地位。這是中美藥企的共同點，而兩者的成長通程不同。從1990 年代到 2010年，一些美國藥企在併購熱潮之後從 Level II發展到 Level I，如Biogen、Celgene。近十年來，美國製藥行業的生物醫學創新轉變為多種來源。過去五年，中國醫藥行業快速發展。2015 年開始的中國藥品政策改革，带來了一些科學的藥品政策，2017年中國参加了國際人用藥品註册技術要求協調會議(ICH)。多家藥企(如恆瑞、信立泰)進入第一等級，研發趨勢從仿製藥轉向創新藥。 結論：美國將繼續主導全球製藥業。中國醫藥產業應抓住機器學習興起的機遇，積極開展國際合作，縮小與美國的差距。 關鍵詞：機器學習、中美比较、醫藥行業、成長性分析