Development of interactive software and AI-based algorithms for the analysis of biomedical data

Abstract

Drastic rise in publications and biomedical data repositories leave modern researchers with the core problem that it is not feasible to consume and understand all available data. Coupled with increasing technical complexity, researchers may thus seek avenues to promote both the accessibility and the impact of their contributions. Progressive web applications (PWAs) may serve researchers in increasing, prolonging, and promoting the relevancy of their work; however producing them introduces a non-research related time constraint, which I herein address via a boilerplate setup for converting aca- demic contributions into PWAs. As a seemingly universal medium, PWAs make otherwise desktop or specialize software accessible even for use in classrooms (e.g. KNIT, SEA). Fur- ther, I’ve made research with expensive hardware requirements (like GPUs) accessible at one’s fingertips (e.g. SCADEN, BED.AI). As the focus is facilitating research not app pro- duction, the development of novel tools and techniques - which may latter be encapsulated in PWAs - was also fundamental for this dissertation (e.g. KNIT, SEA, scGANs, etc). Development of PWAs provide a convenient but time-intensive solution for interfacing with complex, technical, or otherwise cost prohibitive research; however, boilerplate setups lowering time and e↵ort required to manufacture PWAs may result in an influx thereof equally diminishing their value i.e. PWAs could no longer abate publication accessibility and relevancy. Given COVID-19 and a rise in online education, PWAs may also provide a promising avenue to increase scientific literacy via interaction in classrooms. This thesis demonstrates both the need of novel bioinformatic tools (SCADEN, BED.AI, KNIT, SEA, OASIS) in their own right and their increased accessibility when coupled with PWAs.