[Purpose/significance] To achieve the quantitative measurement and early-stage prediction of the original innovation impact of individual scientific papers, this study proposes a method for measuring and predicting original innovation impact. [Method/process] Centered on three core characteristics—original pioneering, original breakthrough, and original leadership—this study constructs a measurement framework for original innovation impact from the perspective of knowledge diffusion, incorporating both short-term and long-term temporal dimensions. On this basis, a prediction method is further developed by integrating semantic information from domain-specific knowledge and
journal-level characteristics. An empirical analysis is conducted using publications in the field of attosecond science
as the research sample. [Result/conclusion] The proposed measurement framework can distinguish papers with high
original innovation impact from highly cited papers, avoiding the simple conflation of impact with citation accumulation. The proposed prediction method can be used for the relative assessment of the potential original innovation impact of newly published papers, providing a quantitative basis for the early identification of research outputs with high original innovation impact.