Co-Loaded PEGylated Nanoliposomes of Bendamustine and Rutin: Formulation, Release Kinetics, and a Hybrid Predictive Modeling Framework

Objectives: Current liposomal drug delivery studies remain largely formulation-specific and descriptive, with limited predictive capability. This study aimed to develop co-loaded nanoliposomes and establish an integrated framework for predictive analysis of drug release. Methods: PEGylated nanoliposomes co-loaded with bendamustine and rutin were prepared using the thin-film hydration method. Physicochemical properties, encapsulation efficiency, and in vitro release were evaluated. An integrated analytical approach combining data augmentation, monotonicity-constrained denoising, Weibull kinetic modeling, and machine learning was applied to characterize and predict release behavior. Results: Co-loaded formulations exhibited higher encapsulation efficiency (up to 77.75%) and distinct release profiles compared to single-drug systems. Weibull modeling adequately described nonlinear release kinetics (R2 ≈ 0.90–0.94). Machine learning enabled within-formulation prediction of later-stage release from early time points (R2 > 0.98; MAE ≈ 0.83–1.00%), although leave-one-formulation-out cross-validation confirmed that cross-formulation generalization remains limited. Reconstructed release curves captured overall formulation-dependent trends, despite variable accuracy in individual kinetic parameters. Conclusions: The proposed hybrid framework enables early prediction of drug release and reveals that curve-level behavior may be approximated without precise parameter estimation, though this reflects parameter compensability rather than robust prediction. This work provides a proof-of-concept framework for analyzing nanoliposomal drug delivery systems.