U of A DataLab Mentors and Capstone Projects

The U of A DataLab with the Data Science Institute and campus partners is developing innovation through student-led capstone projects that tackle real-world challenges in healthcare and accessibility. 

Technology and Accessibility Initiatives:

Graduate students capstone projects from the U of A College of Information Sciences under the supervision of Dr. Greg Chism:

• PubMed Central RAG Agent Chatbot (Mentor: Enrique Noriega).  A state-of-the-art chatbot system is being developed to facilitate learning in biological sciences. The system employs advanced natural language processing and machine learning capabilities to analyze scientific research papers from PubMed and deliver relevant information in a clear, accessible format. The chatbot offers contextual explanations that transform complex scientific concepts into comprehensible material. By combining robust technical architecture with intuitive design, this solution effectively connects academic research with practical understanding, serving as a valuable resource for students, educators, and knowledge seekers.

• Deep Learning Pose Estimation (Mentors: Nirav Merchant, Carlos Lizárraga, Michele Cosi, Anna Dornhaus, Martha Battacharya, and Trevis Wheller). Non-invasive animal behavior monitoring is essential for scientific research, allowing researchers to collect authentic behavioral data in natural habitats. This approach enables precise observations of movement patterns and interactions while preserving environmental integrity. However, implementing reliable tracking systems presents significant technical challenges, especially in managing various environmental factors and changing background conditions that can affect measurement accuracy.

The capstone team collaborated with the Department of Ecology & Evolutionary Biology Dornhaus Social Insect Lab, the Department of Neurosciences Bhattacharya Lab, and the College of Pharmacy, Wheeler Lab at the University of Arizona to develop an integrated cloud-based software solution. This solution streamlines laboratory analysis tasks for pose estimation research communities by integrating DeepLabCut, SLEAP, and DIPLOMAT tools.

• Improving Navigation and Accessibility for the Visually Impaired. (Mentors: Nirav Merchant, Jeff Bishop, Carlos Lizárraga, Jonathan Alberding). The University of Arizona's Campus Accessibility Enhancement Project improves navigation for visually impaired individuals through the use of AI tools. This initiative empowers visually impaired community members to navigate campus independently and confidently. By combining wayfinding technology with detailed spatial information, the project enhances campus mobility—reinforcing the university's commitment to an accessible and equitable environment for all.

In collaboration with the University of Arizona Disability Resource Center, the capstone team explores Large Language Models to generate detailed route descriptions for visually impaired individuals, enhancing the existing University of Arizona Campus Map.

Undergraduate Students Capstone Project from the College of Information Sciences under the supervision of Michael McKisson:

• Analyzing how effective the 2023 rule changes were to MLB attendance and game duration (Mentor: Carlos Lizárraga). This capstone project examines how Major League Baseball's 2023 rule changes affected game attendance and duration. The study analyzes time-series data from a sample of MLB teams, comparing pre-pandemic periods with post-rule implementation periods to assess how these changes influenced fan engagement. The analysis excludes pandemic-affected seasons to maintain data consistency.

Healthcare Innovation Projects  Supported Undergraduate Capstone Projects from the Eller College of Management under the supervision of Gondy Leroy. (Mentor: Gondy Leroy, AI/ML Support: Nirav Merchant, Michele Cosi, Carlos Lizárraga).

• Emergency Department Triage Assessment System  Employing state-of-the-art AI and ML technologies to systematically evaluate and assign priority levels to patients, ensuring optimal resource allocation based on clinical severity and time-sensitive care requirements.

• ER Cost Estimate Chatbot  A sophisticated interface that calculates real-time estimates of emergency care costs by integrating patient insurance details, presenting medical conditions, and comprehensive treatment cost data, enabling patients to make well-informed decisions regarding their medical care expenses.

• Treating Undiagnosed Cases of Rare Diseases   A specialized diagnostic system that leverages advanced analytics to streamline the detection of rare medical conditions by evaluating patient symptoms, health records, and clinical data. The system places specific emphasis on identifying conditions frequently overlooked in conventional medical assessments.

• Adaptive Medical Communication System   A tailored platform that presents medical information through clear visuals and engaging content. This solution adapts medical explanations to meet the specific needs of diverse populations, making healthcare information more accessible while accounting for varied cultural perspectives, educational backgrounds, and age-specific requirements.

The capstone projects highlight the U of A DataLab's dedication to applying AI and data science in areas such as healthcare accessibility and scientific discovery, while guiding students in their professional growth. By bringing together students, faculty, and research labs from various disciplines, these capstone projects contribute to solving complex problems.