Methods and models to control and predict behavior of two dimensional paper networks for diagnostics

Abstract

Medical care is dramatically more efficient when physicians can diagnose their patient's illness to guide appropriate treatment. Currently, there are two classes of diagnostic tests: high-performance gold standard laboratory-based tests that require complex and expensive infrastructure that is not available in clinics, homes, or rural areas; and inexpensive, easy-to-use, rapid lateral flow tests that often have poor sensitivity and specificity. Novel devices that achieve high sensitivity and specificity, while remaining accessible for point-of-care diagnosis in low-resource settings, would dramatically improve heath outcomes around the world. Recent developments toward 2-dimensional paper network (2DPN) diagnostics that combine the sophistication of microfluidics with the simplicity and low cost of lateral flow tests are a promising solution for point-of-care diagnostics. Integration, automation and design of these 2DPN devices would be improved if device engineers could predict how their assays would behave without needing to iterate through dozens of experiments to optimize conditions. These improvements can be realized through the development of techniques to predict and control the behavior of a variety of different components that are used in the assay design process. This project has developed a set of methods and models that are a valuable step toward the goal of fully controlled and "engineerable" assays. First, the process of printing reagents onto nitrocellulose membranes has been characterized--examining and modeling the imbibition of liquids into membranes. Second, techniques were developed to study protein adsorption to nitrocellulose surfaces, and a model was developed to evaluate and compare different rates of this adsorption process. Third, methods were developed to pattern and dry reagents for storage directly within the nitrocellulose membranes, and subsequently rehydrate them in a variety of controlled spatial and temporal distributions in a device. Fourth, a multi-step malaria assay was implemented using solely patterned and dried reagents. Fifth, a computational model of transport and binding of reagents within a single pore was developed to provide insight into the various parameters governing assay signal. Finally, another computational model was built to illustrate the multitude of binding interactions that occur when multivalent analytes and antibodies are used in an assay system.