Background The impact of artificial intelligence combined with advanced techniques is ever-increasing in the biomedical field appearing promising, among others, in chronic kidney disease (CKD) diagnosis. However, existing models are often single-aetiology specific. Proposed here is a pipeline for the development of single models able to distinguish and spatially visualize multiple CKD aetiologies. Methods Acquired were from the Human Urinary Proteome Database the urinary peptide data of 1850 healthy control (HC) and CKD (diabetic kidney disease-DKD, IgA nephropathy-IgAN, vasculitis) participants. The uniform manifold approximation and projection (UMAP) method was coupled to a support vector machine (SVM) algorithm. Binary (DKD, HC) and multiclass (DKD, HC, IgAN, vasculitis) classifications were performed, including or skipping the UMAP step. Last, the pipeline was compared to the current state-of-the-art single-aetiology CKD urinary models. Findings In an independent test set, the developed models (including the UMAP step) achieved 90.35% and 70.13% overall predictive accuracies, respectively, for the binary and the multiclass classifications (96.14% and 85.06%, skipping the UMAP step). Overall, the HC class was distinguished with the highest accuracy. The different classes displayed a tendency to form distinct clusters in the 3D-space based on their disease state. Interpretation Urinary peptide data appear to potentially be an effective basis for CKD aetiology differentiation. The UMAP step may provide a unique visualization advantage capturing the relevant molecular (patho)physiology. Further studies are warranted to validate the pipelines clinical potential in the presented as well as other CKD aetiologies or even other diseases.