Inspiration We were inspired by a painful gap in cardiac care: cardiac rehab is one of the most proven ways to improve recovery and survival, yet too many patients never start or never finish it. The problem is not that the treatment does not work. The problem is that once patients leave the hospital, support often disappears. We built CorVas AI to make recovery feel guided, understandable, and human-supported instead of overwhelming and lonely.
What it does CorVas AI is an AI-powered cardiac recovery companion for patients after a major heart event. It helps patients understand what their doctor said in plain language, stay on track with medications and rehab, monitor recovery progress, and get support before small setbacks become dangerous. It also brings in the human safety net by escalating concerns to family, care teams, or support services when risk increases.
How we built it We built CorVas AI as an interactive product experience across a Next.js and TypeScript app, with a native SwiftUI iOS version for mobile presentation. The product includes live visit transcription, AI-powered medical translation, grounded question answering, rehab planning, medication support, document summaries, and messaging for family and care-circle coordination. For AI, we used an abstraction layer over OpenAI and Anthropic models so the experience could generate plain-language explanations, encouragement, risk-aware nudges, and support responses while staying grounded in visit context and care plans.
Challenges we ran into One challenge was balancing intelligence with safety. In healthcare, it is not enough for AI to sound helpful. It has to stay grounded, avoid inventing diagnoses, and know when to hand off to humans. Another challenge was shaping a hackathon demo that still felt believable end to end, from visit context to rehab adherence to escalation. We also had to move quickly across product design, frontend implementation, mobile experience, AI prompting, and demo storytelling all within hackathon time constraints.
Accomplishments that we're proud of We are proud that CorVas AI feels like more than a chatbot. It is a full recovery journey that connects visits, rehab, medications, documents, support systems, and escalation logic into one experience. We are also proud of the emotional design: the product supports patients through both the “win” moments and the “wall” moments, which is where real dropout happens. Most of all, we are proud that we built something that makes cardiac recovery feel clearer, safer, and more human.
What we learned We learned that adherence is not just a medical problem, it is a design problem. Patients often do not need more information, they need the right explanation, at the right moment, with the right support around them. We also learned that in healthcare AI, trust comes from constraints: grounding responses in real context, being transparent about limits, and escalating to people when needed. Good healthcare AI should support care plans, not replace them.
What's next for CorVas AI Next, we want to make CorVas AI production-ready with deeper personalization, real wearable integration, stronger clinician workflows, and more robust escalation pathways. We also want to expand support features like transportation matching, family involvement, and community-based recovery accountability. Longer term, we see CorVas AI becoming a true longitudinal cardiac care companion that helps patients stay engaged not just for a week, but through the full recovery journey.
Built With
- claude
- openai
Log in or sign up for Devpost to join the conversation.