| Objective: Analyzing the development trend of knowledge diffusion and integration among scientific research results,patents and innovative products in depth can provide an important decision basis for discipline layout,R&D and innovation.However,the existed metrics and data cannot fully reflect the correlation between science,technology and innovation and their change patterns.The purpose of this study was to analyze and study the "science-technology-innovation" linkage through citation networks and content mining by integrating academic papers,patents,and drugs using citation relationships.Methods: Papers,patents,and drugs in the biomedical field were used as research objects to characterize science,technology,and innovation.The citation backtracking method was used to collect FDA-approved drugs,related patents,and academic papers cited by patents,and to study the number distribution,research direction,core institution distribution,and changes over time.The citation network of multi-source data was constructed based on citation relationships,and Gephi was used to map the citation network in different time zones and analyze the evolution of sciencetechnology-innovation nodes and citation relationships in the citation network over time.The strength and changes of science-technology-innovation relationships were further measured by Science Linkage and Technology Linkage indicators;the time lag and changes of science-technology-innovation relationships were measured by Science Cycle Time and Technology Cycle Time.Based on the citation network,the improved BERTopic topic mining model was used for drug,patent,and academic paper topic extraction,and the c-TF-IDF method was used for topic semantic characterization.The word cloud visualization method was used to represent the distribution of important topics in different time slices.Highly similar topic associations were further obtained by inter-topic similarity calculation and thresholding.The "paper-patent-drug" topic association and development paths were depicted by topic categories.The association patterns were classified and explored in depth using a time-series based topic sankey diagram visualization method.Results: The bibliometric analysis indicated that the number of drug-related patents and academic papers in biopharmaceuticals developed rapidly since the 21 st century.Specifically,the types of CNS drugs(17.7%),topical drugs(12.1%),anti-infective drugs(11.9%)and anti-tumor drugs(11.9%)accounted for a larger proportion,with anti-tumor research and CNS being the research areas that have developed at a faster pace in recent years.In terms of core institutional distribution,drugs and patents had an important share of large international pharmaceutical companies.University institutions were the main force of academic publications,and the institutions of scientific research publications were more dispersed compared to the dense distribution of drugs and patents.The citation network analysis constructed by citation relationship found that the citation dependence of drugs on patents is much smaller than that of patents on academic papers,and the citation relationship showed an explosive growth after 2006,and the association was gradually close.The association between drug innovation and patent technology showed a slow fluctuating growth pattern,and the association between patent technology and scientific papers showed a three-stage pattern of low period,slow rise and high-speed development.The average Science Cycle Time(12.6 years)was larger than the Technology Cycle Time(3.1 years),and the rate of increase was slower than the Technology Cycle Time.After further text mining for potential associations,it was found that the biomedical field themes could be divided into three main categories: drug or adverse effect research for specific diseases or indications,drug synthesis research,and drug release pattern research.The latter two categories were the first to combine science and innovation,with the highest degree of "science-technology-innovation" integration.Most of the thematic evolutionary paths in the same category showed the association pattern of "sciencescientific innovation-technological innovation" or "science-scientific innovationscientific innovation".In different categories of thematic associations,the phenomenon of the innovation of science and the phenomenon of the science of innovation were observed.Finally,a time-series-based thematic sankey diagram revealed that there are three patterns of "science-technology-innovation" linkage evolution in the biomedical field: independent development,simple split-fusion and complex synergistic development.Conclusion: This study takes the field of biomedicine as an example and uses multisource data based on citation relationships to confirm that the "science-technologyinnovation" linkage is deepening and becoming an organic whole from the perspective of citation and text mining.The knowledge flow between science,technology and innovation has entered a rapid development stage in the 21 st century,showing the integration process of science,technology and industry and the development pattern of science-technology-innovation linkage,which will provide reference for the evaluation and management of results in the biomedical field. |
PDF Full Text Request