Project Lead: Matthew Louis Mauriello (mlm@udel.edu)
Traditional rehabilitation and gait monitoring often rely on short clinical observations, limiting continuous analysis of patient movement in real-world environments. WalkSignal explores how wearable sensors and machine learning can provide scalable, real-time gait analysis for rehabilitation and mobility research. The project combines wearable force sensors, ML-based gait activity detection, and a mobile application for live sensor data collection and experiment management. Current work focuses on analyzing gait features such as step patterns, activity classification, and movement behavior across different walking conditions. The long-term goal is to support more accessible, data-driven rehabilitation and early mobility impairment detection.
Project Lead: Prerana Khatiwada (preranak@udel.edu)
Misinformation and declining trust in online media continue to be major societal challenges. While traditional fact-checking approaches can help mitigate misinformation, their effectiveness often depends on how transparently and thoughtfully they are implemented, as poorly designed interventions may further reduce trust in news and information systems. Students working on this project are developing user-facing interventions powered by large language models (LLMs) to support more reflective and critical engagement with online news content. These interventions include tools for highlighting potentially misleading rhetoric, encouraging lateral reading behaviors, prompting active learning, and supporting context-aware reflection while users browse news articles. The project also explores human-in-the-loop approaches where community feedback, crowdsourced annotations, and expert input can work alongside AI systems to improve transparency, adaptability, and user trust. In the long term, the research aims to investigate how recommendation algorithms, adaptive interventions, and interactive learning experiences can positively influence browsing behaviors, strengthen media literacy, and support healthier online information ecosystems.
Project Lead: Moath Erqsous (merqsous@udel.edu)
From ELIZA, the earliest psychotherapy chatbot, to today’s large language models (LLMs), AI has long carried the vision of making digital support feel more human. DoveChat continues that vision for adults with ADHD. Built at the intersection of Human-Computer Interaction (HCI), AI, and mental health technology, DoveChat is designed to help individuals navigate focus, overwhelm, organization, and everyday routines with support that is aware of neurodivergent experiences. The project explores how human-centered AI systems can provide accessible and supportive interactions that fit naturally into daily life. We are currently developing DoveChat as a mobile application for Android and iOS platforms to bring personalized AI-powered support into everyday environments.
Project Lead: Aishwarya Chandrasekaran (aishc@udel.edu)
This project explores how contextual factors enhance automated stress prediction and mental health support in naturalistic settings. Traditional wearable sensors often fail to capture the subjective nature of psychological stress when isolated from a user's environment. To address this, the project utilizes multimodal systems that integrate physiological data from smartwatches with behavioral signals—such as facial action units, gaze, and peripheral keystroke and mouse interactions—collected via personal computers. Through "in-the-wild" deployments with information workers and computer science students, this research demonstrates that contextual metadata, including active applications and time of day, frequently outranks raw physiological data in predicting stress. One of the contributions of this research, a hybrid Personal Informatics tool named EmotionStream when deployed in-the-wild helped students reflect on academic stressors like debugging by temporally aligning their automated emotional states with contextual cues. In another contribution, by serializing the multimodal data into structured narratives, we fine-tuned general purpose and domain specific Large Language Models to interpret complex contextual nuances, reframing numerical stress detection into a highly accurate semantic inference problem.
Project Lead: Fatimah Alhassan (alhassan@udel.edu)
This project explores the relationship between energy lifestyles and sedentary behaviors to better understand how everyday habits shape residential energy demand while promoting physical fitness. Students working on this project modify off-the-shelf energy monitoring technology, aggregate data from wearable fitness trackers, and build new in-home interfaces for residential energy customers. The long-term aim is to explore new efficacious feedback and interventions strategies resulting from the integration of these and other data sources, ultimately creating systems that make these relationships visible, interactive, and actionable.
Project Lead: Abhishek Karwankar (karwabhi@udel.edu)
How might music theory and low-cost IoT devices support community members with autism? This project explores the design of technology-mediated musical experiences for children with autism, enabling them to dynamically vary levels of sensory stimulation through interactive, multi-layered music systems. The work brings together tangible musical interfaces, data logging, and interactive visualization to capture and analyze patterns of engagement during music interaction. These insights not only help understand how children respond to different musical structures, but also inform the design of creativity support tools for music composers. Through these tools, composers can interpret interaction and acoustic data to guide their compositional decisions. By integrating generative AI with music-theoretic representations and user interaction data, the system supports data-driven adaptation and personalization of musical experiences. This enables the creation of responsive, individualized music that aligns with each child’s sensory and emotional needs. Ultimately, this work examines how adaptive, human-in-the-loop musical systems can enhance engagement, support socio-emotional development, and empower both therapists and composers in shaping meaningful therapeutic experiences.
Project Lead: Kyle Wang (kylewang@udel.edu)
Recent qualitative studies are increasingly using large-scale online social network data to understand how people interact with technologies and with one another. However, current approaches to dataset generation and analysis are often manual, time-consuming, and difficult to reproduce. This project develops easy-to-use tools to support social media researchers in collecting, filtering, annotating, and interpreting data from platforms such as X/Twitter and Bluesky. A central focus is the use of AI agents to assist with annotation and analysis, with the goal of making large-scale social media research more efficient, transparent, and reproducible. The long-term aim is to build tools that can help a broad range of researchers improve data quality and generate more reliable insights.
Project Lead: Abigail Liu (apliu@udel.edu)
A person may want to change the way their voice sounds for a variety of reasons, and several apps currently offer support for this. However, many of these tools are usually designed or marketed toward one specific group of users. Vocal Goals aims to design a user interface that helps visualize different vocal qualities in a clearer and more useful way. Current work focuses first on gender-affirming vocal training, with plans to expand in the future to adjacent communities such as people practicing public speaking or those receiving voice or speech therapy for vocal cord nodules.
Project Lead: John Aromando (jaro@udel.edu)
Introductory computer science courses, commonly known as CS1 courses, have seen large and growing enrollment, but they also face high dropout rates of around 25%. Instructors work hard to support students, but it can be difficult to scale support across large class sizes while meeting the needs of increasingly diverse learners. Learner-Adaptive, Pedagogical, Interactive Solutions (LAPIS) focuses on finding effective ways to represent CS1 data. LAPIS aims to improve computer science education by making learning more personalized and effective, reducing instructor workload, and helping students succeed in their CS1 courses.
Project Lead: Minji Kong (sensifylab@gmail.com)
Classroom Augmentation examines how interactive technologies can assist teachers in programming classes that use block-based environments such as Scratch. The project studies teaching augmentation systems designed to provide educators with greater awareness of student learning progress and reflective thinking during coding activities. By conducting teacher-centered design research, the project explores concepts including learning analytics dashboards and ambient classroom displays that can support real-time instructional decision making and reduce classroom management overhead. The long-term goal is to develop practical and accessible tools that strengthen computing education and create more supportive learning experiences for both teachers and students.
Project Lead: Sahar Nilipour (sensifylab@gmail.com)
You can tell a lot about a person by their browsing history including whether they are depressed, lonely, or prone to overconsumption of web-based content. Students working on this project are exploring features of web usage data that might be useful in predicting measures of mental health and wellbeing. The long-term aim is to explore how such data can be used in personal informatics tools that provide opportunities to reflect and build wellbeing skills.
Project Lead: Michael Arocho (sensifylab@gmail.com)
The COVID-19 pandemic resulted in the closure of many concert halls, movie theaters, and similar event spaces. Faced with the potential of future global pandemics or other catastrophic events, the main motivation of this project is delivering live concerts and similar cultural experiences simultaneously and synchronously in both physical and virtual formats. The long-term aim of this work is to explore technology-mediated interactions between physical and virtual audiences.
Project Lead: Owen He (sensifylab@gmail.com)
Procedural generation in games is a fascinating way to expand the capacity of development teams to produce new content. However, most of this work focuses on generating maps and digital assets like trees. Students working on this project are developing procedural generation techniques and deep learning models that rapidly generate interesting and dynamics characters. The long-term aim is to explore character-based creativity support tools that integrate well into game development practices and workflows.