An Artificial Intelligence Approach to Develop Tunable Nanoparticulate Delivery Systems for Regenerative MedicineApplications

Background and Aim: Regenerative medicine plays a definitive roleto treat diseases once thought difficult to treat; however, the creationof successful protocols toward widespread clinical application is verychallenging. Efforts have been launched to use artificial intelligence (AI)to accumulate skills and experiences of world-leading researchers toensure quality, low-cost and widespread regenerative therapy. The recentAI applications include: controlled differentiation of iPS cells, cell therapyoutcome prediction, and tissue scaffold biofabrication optimization. Thisstudy highlights the potential of AI to develop tunable nanoparticulatedelivery systems for tissue engineering applications.Methods: Artificial neural networks and genetic algorithm were appliedto fabricate and optimize PLGA nanoparticles with respect to size,zeta potential, polydispersity, loading capacity and release profile. Amechanistic model coupled with genetic algorithm was developed topredict the structural and release characteristics of PLGA/PLLA core-shellbilayer nanoparticles for spatiotemporal control of growth factor release.The nanoparticles were prepared based on the conditions suggested by theAI-based models. Then, the nanoparticles were employed to control coreleaseof VEGF and bFGF and sequential release of PDGF. Based on theendothelial sprouting assay, the angiogenic response of the nanoparticleswas assessed in methacrylated collagen hydrogel with respect to the coreleaseof VEGF and bFGF, sequential release of PDGF, and the releaseof VEGF alone. The novelty of this study is associated with the distinctpotential of the proposed AI-based approach for rate-modulating PLLA/PLGA nanoparticles for tissue engineering applications.Results: The developed AI-based models successfully predicted thenanoparticle fabrication conditions which led us to achieve pre-definedgrowth factor release profiles. The predictability of the AI-based modelswas confirmed by experiments. Notably, as predicted by the AI-basedmodels, the release lag phase of the core-shell nanoparticles wasfollowed by zero-order release kinetics, which is essential for timedelayedrelease of growth factors. A combination of growth factor-loadednanoparticles, which were designed using the AI-model, successfullyprovided different release scenarios of VEGF release only, VEGF, bFGFand PDGF co-release, and sequential release of PDGF. The sproutingangiogenesis, based on the rat aortic ring assay, indicated a significantdifference in angiogenic response between the three release scenarios,suggesting that the designed nanoparticulate delivery system was ableto regulate the predefined growth factor release patterns to promoteangiogenesis in methacrylated collagen hydrogel.Conclusion: The proposed Geno-Neural approach offers great potentialfor the design and optimization of PLLA/PLGA-based nanoparticlesfor preprogramming of various release profiles for tissue engineeringapplications including angiogenesis in hydrogel scaffolds. The hybrid AImodelcan elucidate the release mechanism of growth factors from coreshellnanoparticles. The AI-based model suggests that despite relativelyrapid PLGA core degradation, PLLA shell predominantly controls theoverall release from PLLA/PLGA core-shell nanoparticles. The proposedAI-models can be employed for tuning growth factor release from bilayerPLLA/PLGA nanoparticles.

کلیدواژه ها:

Artificial intelligence ، Nanomedicine ، Controlled release ، Tissue engineering ، Regenerative medicine ، Angiogenesis