We propose a machine learning-aided Engineering Change Order (ECO) delayed prediction model and automated ECO process, based on the random forest algorithm, a “gate + wire” delay model is established, which can conveniently calculate the path delay and accurately predict the change of path delay after ECO. In this paper, considering that different ECO operations will have different impacts on the path, the delay prediction model can adapt to complex ECO operations such as inserting buffers and replacing cell sizes and thresholds, and considering the possible impacts of ECO operations on the front and back cells, the path delay after ECO is finally evaluated through a two-step prediction process.