About one quarter of patients hospitalized with influenza-like illnesses are infected with more than one respiratory virus. Coinfections can lead to more serious outcomes for patients and are more complex to treat than infections with single viruses. Mathematical models can be used to help us understand the dynamics of viral coinfections and optimize treatment. Previous mathematical models of viral coinfections assume a cell-free transmission pathway with virus leaving one cell and traveling to nearby uninfected cells. However, viruses can also tunnel directly from one cell to another, which can affect how coinfecting viruses interact. This project analyzes a system of coupled ordinary differential equations that includes both cell-free and cell-to-cell transmission to better replicate actual viral spread. We measure coinfection duration for combinations of five common respiratory viruses as a function of the amount of cell-to-cell transmission. We find that coinfection duration depends nonlinearly on the cell-to-cell transmission rates, with differing patterns for different coinfecting viruses. This study highlights the importance of considering different transmission modes when modeling viral dynamics.

A key proponent to galaxy evolution is the multiphase gas that surrounds and permeates the galactic disk. Studying this complex gas network allows us to better understand how it regulates the metallicity, structure, and star formation within a galaxy. Within the disk there are small dust grains called polycyclic aromatic hydrocarbons (PAHs). These grains are effective tracers of cold molecular gas and H II regions, as well as production sites for molecular hydrogen which make PAHs excellent probes for studying star formation in galaxies. We use the James Webb Space Telescope (JWST) to study the infrared emission from PAHs throughout the disk of the giant low surface brightness galaxy Malin 1. Compared to high surface brightness galaxies, Malin 1 exhibits less structure and overall dust content. This is potentially hinting at a deficit of cold molecular gas, which is a necessary ingredient for star formation. By mapping out where the dust exists throughout the disk, we can trace areas of stellar formation and gain insight into the properties of this extreme galaxy.

Nanomaterial use continues to rise in biomedical applications, and the need for rapid, accurate, and non-invasive cytotoxicity measurements has become increasingly evident. Existing approaches for evaluating nanomaterial cytotoxicity are often indirect and typically depend on well-plate, salt-based fluorescence assays or complicated microscopy methods. To address these limitations, we introduce FluoAI, a two-stage neural network workflow that directly determines live/dead cells from the nanomaterial’s fluorescence, eliminating the need for additional labeling. The workflow uses two consecutive convolutional neural networks: Mask R-CNN first performs instance segmentation of individual cells from grayscale, single-channel fluorescence images, followed by a DenseNet-121 classifier that assigns alive or dead labels to each segmented cell, achieving performance values of up to 92.0%. In addition to viability classification, FluoAI also performs expert-level analyses of corrected total cell fluorescence (CTCF), mean fluorescence, and cell area, with results showing minimal to no significant differences compared with human measurements of Graphene Quantum Dot (GQD)- and fluorescein dye-treated model cells. Because the entire pipeline is automated, these quantitative fluorescence metrics are generated faster than manual analysis while maintaining comparable accuracy. Overall, this AI pipeline enables non-invasive cytotoxicity assessment and automated in vitro analysis using a conventional fluorescence microscopy setup. As its dataset continues to expand, FluoAI provides a strong foundation for reliable, high-throughput nano-cytotoxicity assays and automated data analysis, ultimately supporting the development of novel and safer nanomedicines.

Graphene quantum dots (GQDs) have gained interest within the bioimaging community due to their biocompatibility and their ability to exhibit near-infrared (NIR) fluorescence suitable for imaging and tracking within biological systems. Creating a simple and reproducible synthesis method for biocompatible NIR-fluorescent GQDs from a variety of precursors remains a critical task. Development of multiple NIR fluorescent GQD structures from a variety of precursors can facilitate their application in multiplex imaging, multianalyte sensing and combination therapy delivery. Herein, we demonstrate the synthesis of 11 distinct GQD structures capable of NIR fluorescence, achieved through a facile microwave-assisted bottom-up carbonization of 11 different materials: ascorbic acid, chitosan, citric acid - urea, dextran, glucose, glucosamine hydrochloride, hyaluronic acid, l-glutamic acid, polyethylene glycol (PEG), sodium cholate, or sodium citrate. All GQD structures exhibit substantial biocompatibility at concentrations up to 2 mg/mL. Internalization of GQDs is observed through their NIR fluorescence, allowing them to be successfully tracked in vitro in HEK-293 cells. This work provides a comprehensive study demonstrating how precursor selection enables versatile synthetic outcomes and NIR-emissive GQDs with distinct physical, chemical, and optical properties relevant to bioimaging. Expanding the precursor range democratizes the use of GQDs in biological applications by providing broader access to structures with tunable NIR emission and surface characteristics.

Oncolytic adenoviruses are promising cancer therapies because they can selectively infect and destroy tumor cells, however their replication in cancer cells is sometimes limited leading to incomplete tumor suppression. Recently, researchers have started to modify viruses to enhance their replication in cancer cells. In this study, we use a system of ordinary differential equations (ODEs) to model tumor growth and compare viral treatment dynamics of a modified oncolytic adenovirus ICVB-1042 and a wild-type adenovirus type 5 (Wt Ad5). The model was fit to experimental allowing us to estimate important model parameters for both viruses: infection rate, infected cell death rate, rate of cell protection by the immune response, rate of cell resistance loss, viral production rate, and viral clearance rate. We found differences in the viral production rates and the clearance rates between the two viruses, providing insight into how genetic modifications have altered viral dynamics. These findings highlight how viral properties determine the effectiveness of oncolytic virus therapy.