Over the past decade, and accelerated due to the COVID-19 pandemic, there has been increasing adoption of decentralized diagnostic testing, where the testing is brought closer to the patient. This trend has largely been fueled by the development of more accurate diagnostic tools and faster and more reliable data connectivity. Decentralized testing has been shown to greatly reduce turnaround times while increasing accessibility to users in remote regions. However, there are challenges that limit its widespread adoption. In this dissertation, we detail the development of tools and technologies to overcome these barriers and expedite the shift towards decentralized diagnostic testing. First, we demonstrate the ability to develop point-of-care (POC) diagnostic tests with performance that rivals that of traditional lab-based methods. We developed a rapid, multiplexed, microfluidic serological test for Lyme disease, a tick-borne disease caused by the Borrelia burgdorferi bacterium. The recommended testing, the standard 2-tiered (STT) approach, is not sensitive for early-stage infections, is labor-intensive, has long turnaround times, and requires the use of two immunoassays (enzyme-linked immunosorbent assay (ELISA) and the Western Blot). We developed a standalone multiplexed sandwich ELISA assay and adapted it to the mChip microfluidic platform. We validated the assay on a rigorously characterized panel of human serum samples and demonstrated that our approach outperforms the STT algorithm on sensitivity while matching its specificity. The form factor of this technology is amenable to use in physician’s offices and urgent care clinics. We also showed exploratory work towards adapting the mChip platform for diagnosis of Zika disease, a mosquito-borne disease caused by the Zika virus, and acute kidney injury, a syndrome characterized by loss of kidney excretory function. Next, we worked on increasing the adoption of rapid diagnostic tests for self- and partner-testing designed to be used in at-home settings. We developed a smartphone application to be used alongside the INSTI Multiplex test for detecting HIV and syphilis infections. The application was designed to provide users with i) instructions on running the test, ii) an automated deep-learning-based image interpretation algorithm to interpret the rapid test results from a smartphone image, iii) a way to save test results and display/share them, and iv) resources for follow-up care. We adopted a user-centered, iterative design process where we worked with a cohort of study participants composed of men who have sex with men and transgender women at high risk for contracting sexually transmitted infections. We then field tested the application with 48 participants over a duration of three months and found high acceptability for the application, both in terms of functionality and helpfulness. Finally, we sought to address a key limitation with deep-learning-based image classification techniques, specifically, the requirement for large numbers of annotated images for training. We developed a deep-learning image interpretation algorithm that could be quickly adapted to new rapid test kits using only a fraction of the images that would otherwise be needed for training the model. The interpretation algorithm followed a three-step, modular process. First, the rapid test kit and the membrane were extracted from the smartphone image. Second, the constituent zones were cropped from the extracted membrane. Finally, a classifier detected the presence or absence of a line in the individual zones. Fast adaptation was demonstrated by adapting a base model, trained using images of a single COVID-19 rapid test kit, to four different rapid test kits, each with different form factors, using few-shot domain adaptation. After training with 20 or fewer images, the classification accuracies of all the adapted models were > 95%. This approach can provide a digital health platform for improved pandemic preparedness and enable quality assurance and linkage to care for consumers operating new LFAs in widespread decentralized settings. Together, these methods provide a suite of tools that could expedite the shift towards decentralized, POC testing.