AmarDoctor

Designing an AI-Driven, Multilingual Digital Health Platform for Underserved Communities

Role End-to-end designer & developer
Focus Human-centered AI, accessibility, conversational interfaces
Output Serving over 12k users in underserved communities

The Friction

Existing digital health tools fail many users due to language barriers, rigid medical terminology, cognitively demanding menu-based navigation, and lack of technology that helps in daily decision making.

Language Barriers

Colloquial symptom expressions are not supported.

Health Literacy Gaps in Everyday Decision-Making

Lack of guidance on which specialist to consult and how to adjust diet, making routine health decisions difficult.

Fragmented Medical History

Paper-based records are often lost, damaged, or hard to maintain over long-term care.

Digital Unfamiliarity

Form-driven interfaces impose high cognitive load.

Design Strategy

Redesigned the interaction from menu navigation to conversational dialogue, allowing users to communicate concerns using natural language.

Prioritized accessibility through voice input and reduced text density.

Interaction Flow

Symptom Assessment

Initial symptom input followed by a dynamic triage flow that adapts to user responses.

Dietary Guidance

Culturally contextual recommendations based on locally available foods.

Medical Documents Digitization

Converts paper records into structured digital data for longitudinal access.

Evaluation & Impact

Among 13,388 users, conversational modules showed the highest engagement.

Engagement Stats

Generated 158,858 interaction events over 28 days.

Module Depth
Dynamic Triage 7.51
Chatbot Module 5.56
Static Forms 1.50

Reflection

"Engagement and trust in health AI are driven by how well interaction models align with users’ language and cognitive patterns".