Patient Proxy - Designing AI-powered patient simulations for nursing students at UMass Boston
2025 - Interaction Design
OVERVIEW
Patient Proxy is a conversational AI simulation platform designed to help nursing students at UMass Boston practice patient communication in realistic clinical scenarios. Students can interact with simulated patients, respond to symptoms and questions, and receive feedback after completing a simulation.
The project was developed in collaboration with the Director of Nursing Simulations at UMass Boston to ensure the experience reflects how clinical simulations are structured and evaluated. By introducing guided patient scenarios and open-ended conversations, Patient Proxy explores how AI-powered simulations could supplement traditional mannequin-based training and give students more opportunities to practice clinical communication.
The final result is a live working prototype demonstrating how conversational AI can support experiential learning in healthcare education.
View Live Demo
Duration
4 Months
My Role
As the UX designer, I led the design of the Patient Proxy experience, shaping how nursing students interact with AI-powered simulated patients during training scenarios.
Key Contributions
• Designed screens in Figma, including the dashboard, patient profile creation, simulation interface, navigation.
• Defined two core interaction flows: guided patient simulations with objectives and open-ended patient conversations
• Collaborated weekly with the Director of Nursing Simulations at UMass Boston to ensure the experience aligned with real clinical training and evaluation practices
• Conducted usability testing with six nursing students, leading to design improvements such as replacing command-based inputs with intuitive action buttons
• Worked closely with an AI engineer who implemented the system and brought the designs to life as a live prototype
THE PROBLEM
Nursing students must develop strong patient communication skills before entering clinical placements. At UMass Boston, simulation training uses medical mannequins and instructor observation, but these sessions are limited by time and availability. As a result, students have few opportunities to practice realistic patient conversations.
THE OPPORTUNITY
Patient Proxy is a conversational AI simulation platform that allows nursing students to practice patient communication in a safe and repeatable learning environment.
Interactive Patient Simulations
Custom Patient Profile
Simulation Feedback
DESIGNING FROM SIMULATION MATERIALS
To design an experience that reflected real clinical training, I held weekly syncs with the Director of Nursing Simulations at UMass Boston. During these meetings, she shared domain knowledge about how nursing simulations are conducted, including how students interact with mannequins, how scenarios are structured, and how instructors evaluate student performance.
These conversations helped me understand how students practice patient communication in simulation environments and informed how the digital experience should mirror the workflows used in UMass Boston’s nursing program.
DESIGNING THE SIMULATION EXPERIENCE
Nursing simulations typically include a scenario description, patient background, objectives, vital signs, and evaluation criteria, which instructors use to assess student performance.
Using these materials as a foundation, I designed the simulation interface to mirror the structure of real training exercises. The system begins with a default patient scenario modeled after an existing nursing simulation, allowing students to practice interactions similar to those used in their coursework. Students can also create or modify patient profiles to explore different communication scenarios.
During interactions, students must balance multiple sources of information, including the patient’s symptoms, emotional state, and health literacy. To support this, I designed a navigation system within the simulation interface that keeps critical information accessible during conversations. Students can quickly reference the scenario overview, patient report, and available clinical actions without losing context during the interaction.
As the project evolved, we also incorporated parameters such as patient emotional state and health literacy levels to better reflect the realities of patient communication. This allowed students to practice adapting their responses to patients who may be anxious, confused, or unfamiliar with medical terminology
Late in development, usability testing with nursing students helped identify areas for improvement. One key change was replacing command-based chat inputs with clear action buttons for tasks such as checking vitals or performing assessments, making the interaction more intuitive during simulations.
Together, these design decisions aimed to create a simulation environment that mirrors real clinical training while giving students a safe space to practice patient communication.
PRODUCT WALKTHROUGH
Simulation Interface
Students interact with the simulated patient through a chat interface designed to mirror natural clinical conversations.
Patient Profile Creation
Students can create or modify patient profiles using structured clinical fields based on real nursing simulation scenarios.
Simulation Feedback
After completing a simulation, students receive feedback and a transcript of their conversation to reflect on their communication.
REFLECTION
Designing Patient Proxy required balancing two complex elements: clinical information and natural patient conversation. Learning how nursing simulations are structured introduced many new clinical details. Weekly conversations with the Director of Nursing Simulations helped me understand how patient profiles are set up, how scenarios are evaluated, and how communication is assessed.
Collaborating with an AI engineer reinforced the importance of discussing technical feasibility early in the design process. By reviewing ideas together before building, we were able to prioritize a focused MVP and avoid designing features that would be difficult to implement.
If I had more time to design this, I would have spent more time learning which patient information is most important for nursing students during simulations. While the patient profile currently mirrors the structure used in existing training materials, deeper conversations with instructors could help identify which fields are most critical for clinical decision-making. This would allow the interface to better prioritize information through clearer hierarchy and reduce unnecessary complexity in the patient profile creation process.