The goal was to put an AI medical assistant in the waiting room before a virtual doctor's appointment. The AI asked the patient questions about their symptoms and logged the answers for the provider to review. The design included both the patient and provider experience. The bot was in the DexCare virtual health platform. My contribution was product design and I worked with DexCare engineering and Gyant to implement the design.
On the left is the patient medical record and demographic information. In the middle is a log of the chat with the patient and on the right is the AI generated diagnostic of the symptoms the patient is experiencing. There is a feedback loop for the doctor to let the team know how useful the AI was. A functional prototype was built in Figma.
The health assistant is a white label chat experience. These images are of design studies of the customer-facing experience with Providence branding. The features displayed are the symptom diagnostic and the FAQ. This bot was built on Microsoft bot framework. Design studies and a functional prototype were built in Figma.
The flow diagram shows the conversation interaction design. Scenarios covered were when a patient asked to speak with a human, when the reason for the visit was not understood, and when serious symptoms are reported. Serious symptoms or alarm phrases are flagged to alert staff of a problem.
Working with a researcher, we conducted on-site customer interviews and learned about pain points and opportunites.
From several focus groups we learned about the persona for the health assistant. We learned what features they expected to find and why they would choose the assitant over another option.
Brainstorming sessions with the team revealed opportunities for personalization. We captured our ideas on a virtual whiteboard, and this helped us collaborate remotely.
