22484 research outputs found
Sort by
Intervention Co-Design: Community Collaboration to Build Trans Affirming Provider Training for College Health and Counseling Centers
No full textTransgender and gender diverse college students (TGDS) experience sexual violence (SV) at rates substantially higher than their cisgender peers. Challengingly, TGDS do not find the health resources available to the general student population in the aftermath of sexual violence accessible or trustworthy. This poster will outline the process we engaged to build our e-learning training Centering gender Affirming Resources in higher Education (CARE). CARE aims to improve college health and counseling center’s (CHC’s) SV prevention and response efforts for TGDS. Using human-centered design activities, we guided a community collaborative comprised of TGDS (n=5), CHC providers (n=4), and community practitioners (n=4) through an intervention mapping process. We conducted online sessions using Zoom and MURAL. Collaborative members identified eight behavioral outcomes necessary to increase CHC’s support of TGD students in the aftermath of SV. By employing interactive approaches to community engagement, we built an intervention that addressed roadblocks to TGDS service utilization
Black Girls as Technosocial Change Agents in a Culturally-Responsive Robotics Camp
No full textOur work seeks to support the agency of Black girls in the classroom by placing them as technosocial change agents through technology creation in CS education. We collaborated with the Manchester Youth Development Center in Pittsburgh. In the past years, we have run culturally responsive robotics camps for Black girls, where we fostered an informal learning environment and joyful learning style, such as encouraging collaboration and idea-sharing, allowing Black girls to collaborate on robot design and building while strengthening community bonding. Our analysis of data collected from the summer of 2022 indicated Black girls could build more liberating identities and social relations and challenge dominant narratives by creating their own robots. Further, we found that placing learners as active changemakers instead of workers who execute tasks for others’ purposes was helpful for culturally responsive teaching
Understanding privacy awareness and mitigation strategies in IoT sensing contexts
No full textIn the IoT era, there is a lack of privacy awareness of sensing contexts that have the ability of sensing devices to perceive and understand the environment or situation in which they are deployed. Lack of privacy awareness can cause privacy and security risks while limiting control over data collection and utilization. To understand people’s privacy awareness, existing literature has reported that owners are typically concerned about their personal data being collected, shared, and analyzed by smart home devices. In contrast, bystanders tend to prioritize their relationships with the owners rather than their data collection privacy. Further, owners have generally been shown to be concerned about their health data, whereas bystanders are concerned about the lack of consent towards data collection.
To understand people's privacy expectations in diverse sensing contexts, we found that indoor location type is a vital factor affecting people's comfort, notification preferences, and allow-deny decisions in various sensing scenarios. However, prior work on users' privacy perceptions has generally investigated space sensing, and there is limited work on privacy perception across space and wearable sensing in the existing literature. Comparing different sensing device types is important because individuals can make more informed decisions about the sensing technology they use (as an owner) or are exposed to (as a bystander), such as how they share or control their personal information. In the context of a voice assistant, both smart speakers as space sensing and smartphones with a voice assistant as wearable sensing utilize a microphone sensor to listen to voice commands and an audio processor to process the audio and respond to the user query. Even though they use the same functionality and sensing modality, we found that users' privacy awareness varies. In addition, we present that a divergence of privacy concerns in the vicinity of space and wearable sensors can lead us to engage in unconventional privacy-preserving behaviors
Multimodal Implantable Microelectrode Arrays for Neuromodulation, Neural Recording, and Neurochemical Sensing with Stable Tissue/Device Interface
No full textNeural Interface technologies have seen remarkable progress, particularly with implantable microelectrode arrays (MEAs), which have become indispensable tools in neuroscience research, neurological diagnostics, and therapeutic interventions. The majority of MEAs have been developed for electrically interfacing with the neurons by recording the electrophysiological activities of neurons and/or electrically stimulating the nervous system.
However, information in the brain propagates via electrical activities and chemical signaling that involves neurotransmitters and neuromodulators, highlighting the need for MEA devices capable of both electrical and chemical interfacing. Multi-modality MEAs capable of recording neural activity (both electrical and chemical) and precisely perturbing neural circuits through neuromodulation techniques are of great value for both basic neuroscience research and clinical applications. Such capabilities are crucial for deciphering the complex interplay between electrical and chemical brain signals, which underpin high-level cognitive functions and various neurological disorders. This thesis explores the design, functionality, and application of multimodal MEAs, emphasizing the significance of their integration into neural tissue for holistic understanding and modulation of brain dynamics. On the front of neuromodulation, we developed a flexible MEA for controlled chemical stimulation and electrophysiological recording. Using conducting polymer coatings, controlled solventless drug delivery in vivo was achieved. The MEAs’ recording and neuromodulation capability was validated with acute in vivo experiments in rat models. We then introduced an antifouling zwitterionic polymer poly(sulfobetaine
v
methacrylate) (PSB) to improve device/tissue integration and functional performance. The DNA aptamer-based electrochemical cocaine sensors with PSB coating demonstrated resistance to biofouling and enzymatic degradation, maintaining high sensitivity in vivo that was previously not achieved. We also developed a dual-mode dopamine (DA) sensing and electrophysiology recording MEAs that can synchronously track tonic dopamine and neuronal dynamics stably for 4-weeks in vivo. We demonstrated the role of the Clock (a circadian master gene) in DA dynamic regulation using this technique. Lastly, utilizing PSB coatings, we optimized the technique to achieve stable DA detection and electrophysiology recording in free-moving ClockΔ19 mutant mice for 4 weeks. In summary, this thesis aims to pave the way for the development of multi-modality MEAs that hold the promise for significant breakthroughs in enhancing our understanding of the brain
Title Page Controlling the Chemistry of Hydrogen at Platinum-Electrolyte Interfaces for Applications in Electrochemical Catalysis
No full textThe hydrogen evolution reaction (HER) is a critical process in electrochemical systems, serving as a method for the generation of dihydrogen (H₂) from water and offering a promising alternative to conventional steam methane reforming for hydrogen production. Platinum (Pt) is widely recognized as the most effective catalyst for the HER, which involves the combination of protons with electrons at the electrode-electrolyte interface. Consequently, extensive research has been devoted to the development of alternative catalysts for the HER, with Pt often serving as the benchmark standard in comparative studies. Benchmarking efforts are typically conducted under a range of conditions, employing either platinum nanoparticles (Pt NPs) or platinum supported on carbon (Pt/C) as reference materials. We have elucidated the degradation mechanisms of Pt NPs for the HER in impure electrolytes, an area that has received comparatively less attention in the context of catalyst stability and performance.
We have taken further advantage of the ability to manipulate the chemistry of hydrogen with Pt electrodes to explore the impact of temperature with applied potential on the electrocatalytic hydrogenation of CO2. Electrocatalysis research has been predominantly conducted near ambient temperatures in aqueous environments or through use of electrochemical devices with ceramic electrolytes that operate at greatly elevated temperatures (exceeding 500 °C). We chose CO₂ reduction as our model reaction due to its prominence and extensive study in both electrochemical and thermal catalysis research. The reaction is particularly intriguing because, while CO₂ reduction proceeds efficiently under thermochemical conditions at moderately elevated temperatures (200-400 °C) with H₂ and CO₂, it remains challenging to efficiently hydrogenate CO2 electrochemically at ambient temperatures. This begs questions about the extent to which thermal activation can be used in tandem with applied potential to accelerate the reaction.
We have developed an electrochemical reactor that incorporates features to simplify data interpretation and facilitate direct comparisons between thermal and electrochemical CO2 hydrogenation. This reactor was first benchmarked for its ability to activate and transport hydrogen between an anode and cathode, followed by measurements of CO2 hydrogenation using a Pt/C anode and copper supported on carbon (Cu/C) as the CO2 hydrogenation catalyst at the cathode
Teacher Mentor Perspectives on the Check and Connect Mentor Program
No full textTruant behavior is exhibited across all grade levels and demographics, with many factors contributing to the behavior. While truancy is a problem in our schools, there is no single approach that has been proven to reduce truant behavior across all social groups. This dissertation gains the perspectives of mentors trained in one specific program aimed at impacting student’s attendance, behavior, and course performance. The program is Check and Connect developed by the University of Minnesota. The aim of the study was to gather mentor perspectives on the Check and Connect mentor training and implementation. Mentor perspectives are important because they provide firsthand accounts about barriers and recommendations for the program.
This qualitative study consisted of interviews conducted with 20 mentors who were trained in Check and Connect. These mentors were trained and implemented the program in two separate schools with very different demographics. School A, was a large sub-urban district with over 2,000 students in grades 9-12. School B, was a small urban district with a student population of around 400 students 9-12.
Interviews provided mentor insights into the training and implementation of the program. Mentors' level of preparedness suggested that the model in which training was received mattered. Mentors in School A received in-person training, while mentors in School B received virtual training. Additionally, while dedicated time for the program varied between schools, mentors' ability to be flexible was essential to building successful working relationships with students.
Recommendations to improve the program included: a) scheduling dedicated time during the school day for mentors to meet with mentees, b) ongoing training and collaboration amongst mentors to discuss challenges and strategies, c) effective training that clearly defines program purpose and the importance of building relationships, and d) identifying strategies to engage students in the program
Trustworthy deep learning on medical images.
No full textDeep learning has achieved substantial advancements across a variety of medical imaging tasks, yet its performance can degrade when models are applied to new clinical settings or data distributions. For instance, a model trained exclusively on clean data might misclassify normal/benign cases as cancerous in the presence of adversarial perturbations. In another example, a model trained with data from one population may underperform on data from another population. This dissertation addresses the need for robust and trustworthy deep learning models in healthcare by developing and evaluating methods to improve model reliability in diverse medical imaging contexts. Key challenges include dealing with limited and potentially inaccurate labels, enhancing model interpretability, integrating clinical knowledge for improved lesion localization and survival prediction, and strengthening model robustness against adversarial attacks, spurious correlations, and data distribution shifts. By tackling these issues, the work contributes to the development of trustworthy deep learning models in medical applications
Public Art and Spaces for Expression: Graffiti, Permission Spaces, and Activating Dialogue
No full textGraffiti in Pittsburgh is criminalized and stigmatized due to decades of rhetoric using broken windows theory. To create spaces to imagine more restorative justice approaches we have researched the history and media framing of the law, and interviewed community stakeholders (writers, attorneys, and community organizations) to identify new possibilities. Pittsburgh has imposed some of the most draconian penalties for non-violent crimes of “vandalism” in the country. One narrator was imprisoned for 2.5-5 years and given $232,000 in restitution. Many do not understand their sentencing and it functions as a “double sentence” impacting employment opportunities and mobility for a lifetime. We focus on the experience of writers and surviving family members, alongside community organizations and attorneys. We identify local and international case studies for alternative models of practice: Paris, Pittsburgh, Bogota
