Play is a fundamental aspect of a child’s development. However, many toys on the market require fine motor skills for children to interact with them, creating barriers for those with varying physical abilities. This highlights the need for accessible play technology, such as adapted toys activated by an accessible switch. Unfortunately, these toys are often expensive and difficult to customize. To fill this gap, we developed the Switch Kit – a low-cost, customizable solution for accessible play. The Switch Kit includes: (1) interactive media created in Scratch; (2) an input device that connects to switches, functioning like a keyboard; and (3) various low-cost, easy-to-make accessible switches. To evaluate the usefulness of the Switch Kit, 10 early intervention providers were given a kit to use with their clients for 4-6 weeks. I hypothesized that a child’s clinical presentation would impact their game play, including the type of game providers selected for their client. To differentiate them, I used quantitative interaction metrics logged from the deployment through Scratch, which tracked measures such as duration of each switch press, the number of switch presses, and games played. Providers used the Switch Kit with 10 children with cerebral palsy, 3 children with Autism Spectrum Disorder, and 7 children with global developmental delay. Children with cerebral palsy had the highest switch press rate, while children with Autism Spectrum Disorder had the lowest. This may indicate that children with ASD are less engaged with the Switch Kit in its current form. This research emphasizes the need for tailored game designs to boost engagement, and offers guidance for providers and families when shaping future game design strategies. 

The use of data visualizations in qualitative research varies widely across disciplines, yet there is little consensus on how these visuals are constructed, evaluated, or effectively integrated. This project employs a data-driven literature review to systematically explore these differences and examine the broader intersection of qualitative research and data visualization. We analyze existing studies from qualitative research journals and evaluate them through the Grammar of Graphics framework. Rather than establishing a rigid standard, this research develops a systematic approach to assess and enhance how qualitative data visualizations are used. By mapping various qualitative fields along a spectrum, we identify key factors—such as disciplinary norms, methodological choices, and technological advancements—that influence the adoption and presentation of data visuals. The produced framework does not merely classify the presence of visualizations but examines their function, effectiveness, and alignment with different epistemological stances. Ultimately, this study aims to improve the clarity, accessibility, and impact of qualitative findings by providing a structured understanding of how data visualizations are utilized. By systematically mapping these variations, this study not only reveals the diverse ways qualitative research engages with visualization but also provides a foundation for more intentional and impactful integration, ensuring that visual tools enhance both the interpretability and communicative power of qualitative findings across disciplines. This study is ongoing, and we will present preliminary findings and their implications on the relationship between qualitative research and data visualization.

Music Emotion Recognition (MER) is a prominent area of research in engineering and data science. With the development of music feature extraction systems, the focus has been selecting relevant features and building predictive models based on them. This study aims to build a small structure that can extract music features, and compute the parameters used in classifying emotions. In this study, Marsyas is used to extract music features, and then LASSO regression model is applied to estimate the valence and arousal with the music features. The calculated valence and arousal are used to classify the music emotion based on Russell's Circumplex Model. This approach provides a view of the whole process of classifying music emotion, from extracting the basic features to calculating the parameters, to classifying the emotion.

The horizontal plane of sound localization is dictated by interaural time difference and interaural level difference, vital for localizing low frequencies (<1.5kHz) and high frequencies (>1.5kHz), respectively. This function is compromised in individuals with unilateral and bilateral hearing loss; however, identical etiologies and severities of hearing loss can have profound differences in sound localization deficits. Head movements improve sound localization in individuals with normal hearing (NH) and hearing loss (HL), but current literature has yet to characterize the nature of these movements. For our experiment, we used virtual reality (VR) to evaluate head movement kinetics during sound localization tasks in individuals with NH and HL. Three 360^o VR environments were developed using MetaQuest and Unity to test an individual’s ability to identify 1) 8 visual targets, 2) 16 sound targets without visual targets, and 3) 32 sound targets with simultaneous visual targets in the horizontal plane. NH individuals (n=10) were administered the VR environments in the order listed above within an audio booth. We used MATLAB to conduct statistical analyses, head movement kinematic analyses and calculate root mean squared error (RMSE). Euler Y head movements in Environment 3 had mean standardized path distance=44.89, peak velocity=164.94^o/second, latency=6.89 second, number of head adjustments=1.78, head movement complexity (polynomial fit order with error <35)=1.95 (std = 20.93, 85.08^o/second, 3.19 seconds, 1.26, 1.14, respectively). The average RMSE of 11.5^o is comparable to similar studies and corroborates our findings. Our additional metrics on head movement establish VR as a viable tool to detect variations in movement patterns. This method quantifies head movements, identifies their potential role in sound localization, and develops accessible VR training for individuals with reduced localization ability.

Atomic-scale computer simulations can be used to predict, explain, and improve the performance of prospective nanoscale transistors and devices, which are key components in modern electronics. In particular, first-principles simulations of electronic transport can predict the conductance of atomic-scale materials by computing the quantum mechanical probability for electrons to move through the material. To simulate open boundary conditions in a finite simulation cell, the electronic transport code TRANSEC uses complex-valued functions known as complex absorbing potentials (CAPs), which simulate electrons flowing into and out of the simulation cell, thereby preventing reflection of electrons from the cell boundaries. The effectiveness of CAPs depends on CAP parameters, such as CAP strength and width, which must be tuned for a given material. In general, wider CAPs usually absorb better, but require more space to accommodate the CAPs themselves, increasing the simulation’s size and hence computing time. The goal of our research has been to optimize the CAP form and volume so as to reduce the CAP’s impact on computing time. We have evaluated CAP forms and widths using a simplified tight-binding model of an electronic transport calculation. Our results indicate that an order 1.8 monomial CAP is a highly efficient CAP form, and appears to compare favorably against previously used Gaussian CAPs. Our finding of the optimal monomial range is in broad agreement with previous findings of Seideman & Miller [J. Chem. Phys. 96, 4412]. We have also reproduced these results with TRANSEC, showing that monomial CAPs of monomial order between 1.5 and 2.0 may absorb electrons effectively even for a narrow CAP width, potentially reducing computing time by 25% to 50%. We have confirmed that monomial CAPs of order 1.8 can be tuned successfully for several different nanoscale structures, and can reduce computing time. 

The project aims to design a multi-modal sensor network with VLF antennas will be implemented to model the ionospheric D-region in real-time. In consideration of not having ground truth data, such a network will address the ill-posed problem of inverting with robust regularization techniques. High-data-rate acquisition, high-data-rate processing, and dynamically adaptable auto-tuning will be included in our design. Drawing on experience with the NeSSI, modularity and a digital bus for centrally processed, real-time processing will be part of a standardized, modular sensor network that will be designed. The D-region, an upper atmospheric dusty plasma, controls radio wave propagation via fluctuations in charge. Numerical simulations in our work simulate such occurrences as HF to UHF range radar echoes, validated through experiments in radar labs. Ionospheric instabilities in occurrences such as SAPS events generated through space weather result in GPS and Starlink communications outages. 3D electrostatic fluid and gyrokinetic equations are included in our model, which is significant for describing such instabilities. Real-time observation, predictive maintenance, and reliability in communications networks are enhanced through such studies.