基于机器学习和熵权法的医药领域高价值专利识别研究  

Research on the Identification of High-Value Patents in the Pharmaceutical Field Based on Machine Learning and Entropy Weight Method

作　　者：肖宇锋[1] 杨雪梅 黄雅兰 唐小利[1] XIAO Yufeng;YANG Xuemei;HUANG Yalan;TANG Xiaoli(Institute of Medical Information/Medical Library,Chinese Academy of Medical Sciences&Peking Union Medical College,Beijing 100005)

机构地区：[1]中国医学科学院北京协和医学院医学信息研究所/图书馆,北京100005

出　　处：《中国发明与专利》2025年第4期13-21,共9页China Invention & Patent

基　　金：中国医学科学院医学与健康科技创新工程(重大协同创新项目)基金项目“生物医学文献信息保障与集成服务平台”(编号:2021-I2M-1-033)研究成果。

摘　　要：[目的/意义]通过随机森林与熵权TOPSIS结合的混合算法识别高校和科研机构高价值专利,以期提升医药技术的转化效率,为新质生产力培育和发展提供支持。[方法/过程]本文将“保护内容”纳入医药领域的专利价值评价指标,对医药领域专利进行分类的同时,解决目前研究中缺乏具有医药领域特色的大批量专利定量评价指标体系的问题。通过粗糙集算法进行指标约简,根据准确率、召回率等评价指标选取出评估效果最好的算法为随机森林算法,最后结合熵权TOPSIS算法识别出高价值专利,避免采用单一算法可能造成的结果偏倚。[结果/结论]初步形成适用于医药领域的高价值专利评价指标体系,筛选出高校和科研机构的高价值专利4020项,并结合定性和定量分析,证明所用方法有效。实证研究仅针对肿瘤领域的专利进行了探索,后续可进一步拓展至更多领域以验证该方法的适用性。[Purpose/Significance]To identify high-value patents of colleges and research institutions through a hybrid algorithm combining Random Forest with the Entropy Weight TOPSIS method,with the aim of enhancing the transformation efficiency of pharmaceutical technologies and providing support for the cultivation and development of new quality productive forces.[Method/Process]The“protection content”of the patent is incorporated into the patent value evaluation indicators for the pharmaceutical field,categorizing medical patents while addressing the issue of the lack of evaluation indicator system applicable to the batch patents in the pharmaceutical field.The indicators are then simplified using the rough set algorithm.Based on evaluation metrics such as accuracy and recall rate,the Random Forest algorithm is selected as the most effective for assessment.Finally,high-value patents are identified by integrating the algorithm with the Entropy Weight TOPSIS method.[Result/Conclusion]An evaluation indicator system suitable for the identification of convertible patents in the pharmaceutical field is formed.Combined with the artificial algorithm Entropy Weight TOPSIS,4020 high-value patents from universities and research institutes are screened.The proposed methods are proven effective through a combination of qualitative and quantitative analysis.The empirical research has only explored patents in the field of oncology,and in the future,it can be further expanded to other fields to verify the applicability of the method.

关 键 词：高价值专利 专利转化 机器学习 熵权TOPSIS 医药领域 

分 类 号：R73[医药卫生—肿瘤] G306[医药卫生—临床医学]