RII Track-4: NSF: An Integrated Multiphysics Machine Learning Modeling and Experimental Framework for Optimizing Micro-Needle Patches

点击翻译按钮获取中文摘要

Microneedle patches (MNPs) have provided a solution for different problems associated with needle injection in children and adults such as needle phobia, pain, infection, and even the requirement for a specialist. MNPs deliver a local, pain-free, safe, high-efficiency, and cost-effective way for drug and vaccine delivery. The small needles on MNPs are barely visible to the naked eye. Therefore, manufacturing such products with such details requires state-of-the-art techniques. Among different methods, one of the most efficient techniques is additive manufacturing (3D printing), which itself is a complex process and is controlled by various environmental and physical parameters. Controlling and optimizing all factors at different stages of production is vital for achieving a target design of MNPs. The design of the final product, consequently, controls its mechanical properties. Since the process of optimizing all the parameters involved is computationally very expensive, a machine learning technique will be applied in this project. To test the hypothesis, the 3D printing equipment, which is uniquely available at Stanford University, will be advanced. The proposed research and the associated partnerships will pave the way for developing more efficient MNPs by shedding light on the underlying phenomena and integration of theory and experiments.This Research Infrastructure Improvement Track-4 EPSCoR Research Fellows (RII Track-4) project would provide a fellowship to an Assistant Professor and a graduate student University of Wyoming (UW). In the field of 3D printing, one of the popular printing techniques used in fabricating MNPs is continuous liquid interface production (CLIP), which is a category of vat polymerization technique. Aside from the variables involved in the manufacturing device, processes, and materials, the whole process occurs in a multiphysics environment, which has made the development of computational modeling complicated and time-demanding. All these variabilities can lead to insufficient repeatability, uncertainty, and inconsistency between the produced MNPs, and what is considered the target model, and often the targeted structure is not produced. Reducing uncertainty is one of the prominent problems in MNPs fabrication which we aim to study by integrating both theoretical and experimental tests within the machine-learning framework. We hypothesize that more accurate and effective MNPs, in terms of mechanical stability, can be produced when a large number of scenarios are tested in a closed-loop framework, and it also can help reduce the cost and required time significantly. The proposed research involves a real-time and supervised process that can potentially transform our understanding of the underlying parameters of efficiency and how such effects control the performances.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.