Semiconductor testing requires making precise electrical contact with chips at increasingly small scales, a technique known as fine-pitch probing. Traditional probing methods struggle at very small pitches due to mechanical wear and contact reliability issues. To address this, this project explores gallium-based liquid metal (LM) as a probing material. Its room-temperature fluidity allows it to conform to small contact pads without mechanical limitations. However, a key challenge is the native oxide layer that forms on LM surfaces, which disrupts consistent deposition and wetting. This project investigates how oxide layers influence LM wettability and contact behavior on microfabricated test structures. A test site of microscale pillar arrays with varying pitch is being fabricated as controlled contact interfaces. The focus is on understanding how hydrogen plasma treatment affects the oxide layer and the LM contact angle. Oxidized LM droplets are exposed to hydrogen plasma under varying treatment times and source distances, with droplet behavior captured by video and quantified through frame-by-frame contact angle analysis. This work is expected to reveal how plasma exposure conditions relate to oxide reduction and wettability control. The significance lies in establishing a controllable method for managing LM surface behavior, essential for achieving consistent fine-pitch probing and developing cleaning methods for LM residues on substrates.

Stroke is a leading cause of long-term disability in the United States, with 25% of adults experiencing stroke in their lives. Motor impairments post-stroke are commonly assessed using macro-scale movement metrics. However, similar movements stem from separate neuromuscular control impairments, limiting the effectiveness of current evaluation methods and potentially convoluting rehabilitation protocols. Stroke is fundamentally a control disorder rather than movement alone. Hence, neuromuscular analysis on the patient’s control impairments is critical for personalized treatment. This work presents such neuromuscular insight through a bisensory device for concurrent muscle deformation and neural drive analysis. By integrating mechanical and electrophysiological sensing, this device provides direct insight into neuromuscular control mechanisms and enables classification between different control impairments, like tissue spasticity and reduced neural drive. The device exhibits an SNR of 31.26 dB in deformation, alongside neural signal acquisition with a RMS of 20 μV and a power amplification of four. This first quarter consisted of sequential experiments for device validation, optimization, and bisensory trend analysis. Ongoing device characterization and elaborate clinical testing is projected to further evaluate device performance. These findings establish the feasibility of a neuromuscular sensing device to complement existing clinical tools with real-time monitoring to personalize rehabilitation.

Predictive algorithms and bioinformatic analyses are valued for many biological problems, including drug-effect prediction and patient characterization. However, they often generate large data tables that are hard to interpret. In prior work using protein-protein interaction (PPI) networks to model drug relationships to endometriosis, we observed that predicted drug effects were associated with hundreds of thousands gene ontology (GO) terms, requiring structured summaries of biological themes. To address this, we implemented a computational pipeline to consolidate predictions into broader functional categories. Building on this framework, we explored how pathway-based drug groupings corresponded to real-world patient characteristics by analyzing covariates from an electronic health record (EHR) study of anti-diabetic drugs. Patients taking these medications were associated with many diagnoses and co-prescribed medications, making interpretation similarly challenging and motivating more structured categories. We leveraged the previously developed workflow and adapted it to identify trends in patient covariates. We conducted exploratory analyses of diagnosis data and grouped related conditions into broader disease categories. We developed a framework for comparing covariates across cohorts, enabling more structured interpretation of EHR data. This work provides insight into clinical characteristics associated with each drug group and supports more interpretable results for bioinformatics approaches.

The goal of this study is to develop an automated framework for calibrating high-fidelity microwave ablation (MWA) models by solving the inverse problem of recovering biophysical tissue parameters from experimental thermometry. MWA modeling requires accurate, temperature-dependent tissue parameters, yet studies frequently rely on generic literature values rather than experimental calibration. While sensitivity analyses have identified high-impact parameters, the inverse problem, recovering coefficients from measurements, remains under-investigated. A 2D axisymmetric EM–bioheat MWA model (2.45 GHz, COMSOL 6.2) incorporating temperature-dependent dielectric properties, thermal conductivity, and latent heat was controlled via MATLAB LiveLink. Using a penalty method for physical bounds, Nelder-Mead optimization minimized the RMSE between simulated and experimental temperature traces. Cable efficiency was the primary free parameter (bounds: [0.30, 0.80], initial: 0.52). The optimizer converged in 22 evaluations, calibrating cable efficiency to 0.7378 and reducing RMSE from 9.70°C to 4.61°C (52.5% improvement). A residual peak temperature gap (simulated: 67.0°C vs. ex vivo: 73.3°C) suggests additional parameters are needed. Future work will expand the parameter set to include dielectric drop-off and vaporization window coefficients, using cross-validation to prevent overfitting. This framework demonstrates the potential for experimental thermometry to enhance the predictability and safety of clinical tumor ablation protocols.

Protein-polymer bioconjugation is a promising strategy for enhancing enzyme stability in medical and biotechnological applications, yet synthetic polymers raise concerns regarding biocompatibility and sustainability. This project investigates whether the conjugation of lysozyme and horseradish peroxidase (HRP) to natural and semi-synthetic polymers can improve thermodynamic, thermal, and chemical stability compared to traditional synthetic polymers. BCA assays were used to establish baseline enzyme concentrations. The proteins were then conjugated to their respective polymers by EDC/NHS coupling, with successful conjugation evaluated using SDS-PAGE. Differential scanning fluorimetry (DSF) and 14-day kinetic absorbance-based activity assays were utilized to compare protein stability of the native protein and conjugates. By integrating sustainable polymer chemistry, this work aims to efficiently synthesize biocompatible protein-polymer conjugates that retain catalytic activity while exhibiting enhanced resistance to denaturation.