Human activity recognition is the task of identifying a person’s movement from sensors in a wearable device, such as a smartphone, smartwatch, or a medical-grade device. A great method for this task is machine learning, which is the study of algorithms that learn and improve on their own with the help of massive amounts of useful data. These classification models can accurately classify activities with the time-series data from accelerometers and gyroscopes. A significant way to improve the accuracy of these machine learning models is preprocessing the data, essentially augmenting data to make the identification of each activity, or class, easier for the model. <br/>On this topic, this paper explains the design of SigNorm, a new web application which lets users conveniently transform time-series data and view the effects of those transformations in a code-free, browser-based user interface. The second and final section explains my take on a human activity recognition problem, which involves comparing a preprocessed dataset to an un-augmented one, and comparing the differences in accuracy using a one-dimensional convolutional neural network to make classifications.

Plant-made virus-like particles (VLPs), composed of HIV-1 Gag and deconstructed gp41 proteins, have been shown to be safe and immunogenic in mice. Here, we report the successful production of HIV-1 Gag/dgp41 VLPs in Nicotiana benthamiana, using an enhanced geminivirus-based expression vector. This novel vector results in unique expression kinetics, with peak protein accumulation and minimal necrosis achieved on day 4 post-infiltration. In comparing various purification strategies, it was determined that a 20% ammonium sulfate precipitation is an effective and efficient method for removing plant proteins and purifying the recombinant VLPs of interest. If further purification is required, this may be achieved through ultracentrifugation. VLPs are a useful platform for a variety of biomedical applications and developing the technology to efficiently produce VLPs in the plant expression system is of critical importance.

Influenza virus A (IVA) poses a serious threat to human health, killing over 25,000 Americans in the 2022 flu season alone. In the past 10 years, vaccine efficacy has varied significantly, ranging from 20-60% each season. Because IVA is subject to high antigenic shift and strain cocirculation, more effective IVA vaccines are needed to reduce the incidence of disease. Herein we report the production of a recombinant immune complex (RIC) vaccine “4xM2e” in Nicotiana benthamiana plants using agroinfiltration for use as a potential universal IVA vaccine candidate. RICs fuse antigen to the C-terminus of an immunoglobulin heavy chain with an epitope tag cognate to the antibody, promoting immune complex formation to increase immunogenicity. IVA matrix protein 2 ectodomain (M2e) is selected to serve as vaccine antigen for its high sequence conservation, as only a small number of minor mutations have occurred since its discovery in 1981 in the human sequence. However, there is some divergence in zoonotic IVA strains, and to account for this, we designed a combination of human consensus, swine, and avian M2e variants, 4xM2e. This was fused to the C terminus of the RIC platform to improve M2e immunogenicity and IVA strain coverage. The 4xM2e RIC was produced in N. benthamiana and verified with SDS-PAGE and Western blot assays, along with an analysis of complex formation and the potential for complement activation via complement C1q ELISA. With this work, we demonstrate the potential of RICs and plant-expression systems to generate universal IVA vaccine candidates.