Determining protein-protein interactions (PPI) is crucial in understanding cell structure and biological functions. Accurate prediction of these interactions remains a challenge. In this study, we proposed a solution approach that combines the Hilbert transform applied to evolutionary information for feature extraction and the Deep Forest (DF) classification model for the prediction step. The Hilbert transform enhances the evolutionary information stored in protein sequences. DF, additionally, is a powerful ensemble classification model that improves prediction performance. The proposed method was evaluated on the standard Yeast dataset and compared with other existing methods, showing high effectiveness in PPI prediction. Datasets and source code are available at https://github.com/mxuanvan02/hilb_extract_feat.git