Purpose: The citing of patents is an indicator of the transfer of technical knowledge and the influence of one technology on subsequent developments. The objective of this paper is to construct a predictive model based on patent and paper citation network to evaluate the potential for technology transfer of patents  Methodology: This paper employs a binary classification algorithm from machine learning, specifically the decision tree algorithm, as its methodology. This has facilitated the creation of a predictive index system comprising 16 structured indicators across three dimensions: the degree of technological innovation, the quality of patent document, and the extent of intellectual property rights (IPR) protection. In order to construct the training sample set, patent transfer data for graphene sensors were employed as proxy variables for technology transfer potential. Findings: The findings of the study indicate that the dynamic characteristics of the citation network are a significant factor in the assessment of the potential for technology transfer. The most significant indicator affecting the transformation value of patents is inward degree centrality. Moreover, the results demonstrate that additional evaluation indicators, namely outward proximity centrality, family size, and self-citation frequency, are also of significance. The patent technology transfer potential prediction model presented in this paper, which is based on citation network indicators and a decision tree algorithm, exhibits a high degree of predictive accuracy. Value: The value of this research lies in its novel perspective and method for patent valuation and technology transfer strategies. Furthermore, it offers a more comprehensive reflection of the inheritance and flow of patented technology by integrating the citation relationships between patents and scholarly papers. This makes it a valuable contribution to the understanding and promotion of the technology transfer process.