Human-Centered AI Research Boosts Patient Education Chatbot

​CASE STUDY

The Challenge 

Santovia Health provides patient education services as part of employee health benefits, offering telehealth consultations, physical therapy content, and educational materials to help companies support their employees’ health needs. 

The Roux Institute and the Institute for Experiential AI (EAI) worked with Santovia to conduct user research and create an AI chatbot for Santovia’s service offerings so patients could get personalized health information and discuss symptoms – an opportunity that is often lacking in current care continuums. 

The Partnership 

Dr. Mahsan Nourani from the Roux Institute collaborated with the Institute for Experiential AI (EAI) to lead user-centered research, while EAI built large language and retrieval-augmented generation models for a health information chatbot. The user research was a crucial step to include user perspectives in the product design and implementation life cycle—from the very start.  

Dr. Nourani used her expertise in human-centered AI to design and run a usability study with patients who had experienced medical conditions. The study garnered feedback and helped Santovia learn how people would interact with the chatbot and what value it would bring to their customers. Study insights also proved instrumental in guiding the company’s strategic decision-making regarding future investment in this direction and determining the optimal development approach. 

The Outcome 

Dr. Nourani and team delivered a comprehensive customer discovery and user experience report analyzing how people reacted to the chatbot’s responses with respect to their specific health conditions, including assessment of response quality and relevance. As an expert in human-centered AI, Dr. Nourani took a responsible AI approach, given the high-stakes nature of healthcare applications. Her analysis included recommendations on incorporating guardrails like having the model express uncertainty when appropriate, including disclaimers in responses, maintaining neutrality, and encouraging users to consult with their medical care team. The team also communicated and discussed ethical considerations around how to responsibly incentivize users to engage with the service without pushing them to use it, given the high stakes scenario.  

These customer discovery approaches were critical for de-risking the product from company, legal, and user perspectives, ensuring safe usage and avoiding user misunderstanding during adoption. The report included key findings and recommendations for future improvements that Santovia could make in its product lifecycle.

Interested in learning more about human-centered AI services? Get in touch.