DIGITAL HEALTH

AI in Medicine Fellowship

Led development of AI-powered healthcare solutions at the University of Kentucky Institute of Bioinformatics, including chatbots, monitoring systems, and prediction algorithms.

Rasa FrameworkAzure MLPythonHealth TechEmpathetic AI

The Challenge

Healthcare AI applications often fail to address the emotional and psychological needs of patients. Our challenge was to create AI-powered solutions that feel supportive, accessible, and genuinely helpful while maintaining clinical accuracy.

Sensitive Conversations

Designing for intimate health discussions requires exceptional empathy

Trust Building

Users needed to feel safe sharing personal health information

AI Limitations

Balancing AI capabilities with clear limitations and human oversight

AI Health Companion

Empathetic Interaction Design

Key Projects

Menopause Symptom Chatbot

Built and deployed Rasa-based chatbot focusing on empathetic interaction design and accessible user interfaces.

Remote Glucose Monitoring

Led development with extensive user research with diabetic patients to understand pain points and workflow needs.

Migraine Prediction Algorithm

Developed Azure-based prediction system with focus on clear, actionable user notifications and anxiety-reducing interface design.

Clinical Dashboards

Created intuitive dashboards using Python and R to help doctors interpret complex pathology data.

Conversation Design Principles

Empathetic Responses

Crafted responses that acknowledge user emotions and validate their experiences without being patronizing.

Progressive Disclosure

Gradually introduced more detailed questions as users became comfortable, avoiding overwhelming initial interactions.

Privacy First

Clear communication about data handling and multiple options for users to control their privacy settings.

AI

Hi there! I'm here to help you track and understand your menopause journey. How are you feeling today?

I've been having some hot flashes lately...

AI

I understand that can be really uncomfortable. Would you like to track when these happen so we can look for patterns together?