The asphalt mixture is a complex composite material consisting of asphalt, aggregates, mineral powder, additives, and other components. Its intricate composition poses challenges for traditional design methods in optimizing pavement performance. Moreover, the design and construction processes are accompanied by significant uncertainty and variability, greatly hindering design sustainability due to the vast need for repeated laboratory experiments. Therefore, this study aims to regrade traditional design concepts utilizing digital technology. To this end, two databases were established to create a data-driven system. The Base Database comprises 100 specimens produced through randomized trials and digitized using industrial computed tomography (ICT) techniques. The History Database contains 257 specimens obtained from collating historical data. Afterward, one complete parameter system was constructed, referring to design, internal structure, and pavement performance. Besides, the preprocessing methods of ICT images were enhanced based on accuracy, robustness, and computational efficiency. Finally, a stepwise regression was applied to establish causation among three layers of parameters. The research results confirmed the feasibility of achieving sustainable design by digitizing the whole process. The highest variation coefficient observed in structural and performance layers reached approximately 80%, while the final model’s determination coefficient (R2) in prediction analysis exhibited reasonably high accuracy at 0.920. Therefore, the effectiveness of two critical causal chain analyses—variable traceability and performance prediction—was also validated.