<p>Predicting reuse patterns of novel technologies is crucial for understanding technology diffusion and identifying high-value technologies. This study defines the combination of technology components (i.e., the International Patent Classification codes, IPCs) as "technology" and the invention patents that apply these technologies within the first year of their emergence as "early inventions". A framework was proposed to predict reuse patterns of novel technologies based on the characteristics of technology components and early inventions. A shape-based clustering analysis was conducted on the reuse trajectories of novel technologies that emerged from 1995 to 2020 to identify potential technology reuse patterns. Subsequently, the differences between technologies with diverse reuse patterns were compared. Finally, ten machine learning algorithms were employed to predict the reuse patterns of novel technologies. The findings indicate that the trajectories of 39,376 novel technologies exhibit four types of reuse patterns: s-shaped trajectory pattern, fleeting trajectory pattern, linear trajectory pattern, and exponential trajectory pattern. Furthermore, technologies with diverse reuse patterns substantially differ in their accessibility and similarity of technology components, as well as in the applicability and attention of early inventions. The Gradient Boosting Decision Tree algorithm (GBDT) yields the best performance in predicting the reuse patterns of novel technologies, and the applicability of early inventions serves as the most significant predictor.</p>