Aging is accompanied by widespread molecular changes throughout the human body. Identifying age-associated biomarkers may provide insight into mechanisms underlying age-related disease. Chronic low-grade inflammation, also known as inflammaging, is a recognized hallmark of aging; however, limitations on human research have made it difficult to directly understand inflammaging processes in humans. While inflammaging is recognized as a systemic feature of aging, it remains unclear how inflammatory gene expression patterns differ between males and females across human tissues. Sex-specific differences in immune regulation have been observed in both human and murine studies, yet the extent to which aging-associated inflammatory pathways exhibit sexually dependent transcriptional signatures in humans is not well defined. Various murine studies have revealed of sex-dependent effects of gene expression changes related to aging, such as elevated levels of IL-6r, IL-18, and CRP in women, all of which are associated with cardiovascular and metabolic risk. Consequently, there is an emergent need for identification of sex-related inflammaging biomarkers. Factors such as ethical constraints and tissue collection pose further barriers in identifying these markers; these restraints can be circumvented through the use of open source gene expression data, such as the GTEx data used in this project. We hypothesize that aging is positively associated with systemic  and tissue-dependent changes in expression, with inflammatory pathways exhibiting sex-dependent transcriptional patterns that contribute to distinct trajectories of aging between males and females.

Focal epilepsy is associated with lesions confined to one cerebral hemisphere and may disrupt normal cortical maturation in childhood, though the timing and anatomical specificity of this disruption remain unclear. We examined inter-hemispheric differences in cortical function using resting motor threshold (RMT), the minimum stimulation intensity required to elicit a motor response in a resting muscle and an index of cortical excitability measured via transcranial magnetic stimulation (TMS). 

RMT data were analyzed from 154 individuals with focal epilepsy (ages 1-23 years). Generalized additive models characterized nonlinear age-related trends in RMT, motivating segmented (change-point) regression to estimate the age of maturational stabilization. Within-subject comparisons were performed between epileptic and healthy hemispheres. To assess anatomical specificity, regression models examined lesion-related RMT deviations and compared stabilization ages between corticospinal tract and frontal lobe lesions.

In healthy hemispheres, RMT stabilized at 10.81 years (95% CI [10.49, 11.12]) versus 13.68 years (95% CI [13.45, 13.91]) in epileptic hemispheres, reflecting a mean delay of 2.87 years (95% CI [2.48, 3.26], p < .001). Delays were greater in corticospinal tract lesions (4.16 years, 95% CI [3.98, 4.65], p < .001) than frontal lobe lesions (2.03 years, 95% CI [1.55, 2.52], p < .001). Corticospinal tract lesions were also associated with larger RMT deviations and later stabilization.

These findings demonstrate lesion-specific delays in cortical maturation, with particular vulnerability of the corticospinal tract. Improved characterization of maturation timing may inform earlier, targeted interventions to mitigate long-term motor and neurodevelopmental impairments.

Student engagement, active participation, and collaborative thinking are often absent from the traditional college lecture model, especially in a mathematical or technical course. Such one-way lectures also do little to assure students that they are learning at a pace consistent with class expectations, or that their ideas and participation are welcomed and valued. In this study, we compare the traditional lecture model with collaborative concept mapping, an interactive activity where students work together to create a node-link diagram that organizes ideas and visually demonstrates relationships between them. We collected a total of 60 survey responses from 19 UMass students in a statistics course, which included quiz questions about course material and opinion-based questions assessing perceived sense of belonging, self-efficacy, and exerted effort in class. Students were asked to complete the survey after four class sessions, with two taught primarily through lecture and two through a collaborative concept mapping exercise. The data will be analyzed through mixed models to measure the association between concept mapping and higher academic performance, as well as stronger senses of belonging and self-efficacy. The findings will help to improve teaching techniques and course designs of college-level statistics courses, as well as help future students in statistics courses learn more effectively and feel more comfortable and confident in class.

Autonomous agents operating in shared environments face a fundamental tension between achieving their objectives and concealing their intentions from others who may exploit that knowledge. This research develops a theoretical framework and computational model to study adversarial dynamics in multi-agent systems where agents must balance goal achievement against strategic information management.

We model a sequential game between two classes of agents with asymmetric roles and capabilities. One agent class possesses private objectives and must take observable actions over time to achieve them, while simultaneously minimizing the information revealed through those actions. The opposing agent class observes only the environmental signals generated by these actions and attempts to infer the first agent's hidden objectives to gain competitive advantage. Crucially, the information available for inference is generated endogenously through the goal-directed agent's own behavior, creating a dynamic feedback loop between action selection and information leakage.

Our framework synthesizes insights from classical resource allocation games, signaling theory, and multi-agent reinforcement learning. We introduce temporal constraints requiring agents to achieve objectives within fixed horizons, creating non-trivial strategic tradeoffs.

We validate our theoretical framework through computational experiments using policy gradient methods, demonstrating emergent equilibrium behaviors including strategic timing of actions, adaptive belief formation, and an evolving detection-obfuscation arms race. Our results contribute to understanding strategic behavior in competitive multi-agent systems and demonstrate how reinforcement learning can illuminate complex dynamics arising from information asymmetry and adversarial interaction.