💡 Inspiration Navigating healthcare systems in Bangladesh can often feel overwhelming, especially when trying to find specialized doctors, specific medical tests, or nearby facilities. Families frequently rely on word-of-mouth recommendations or navigate fragmented online directories that lack localized filters or native language support. We wanted to build a simple, accessible solution—HealthMatch-BD—that bridges this gap by allowing anyone in Dhaka to search for medical tests, filter by gender or area, and communicate naturally using bilingual voice inputs in Bangla or English.
🛠️ How We Built It HealthMatch-BD was built using a modern full-stack Web stack designed for high performance and accessibility:
Frontend: Built with Next.js (App Router) and styled using Tailwind CSS with a deep indigo theme palette (#1E1B4B).
Backend & Database: Powered by Supabase and PostgreSQL to manage doctor profiles, hospital networks, available medical tests, and geographic data across Dhaka.
Voice Search Integration: Uses the native Web Speech API configured explicitly with bn-BD locale enforcement for seamless Bengali speech recognition, enabling hands-free voice querying alongside standard text search.
đźš§ Challenges We Faced Bilingual Voice Input Handling: Integrating continuous speech-to-text for Bengali while preventing string concatenation errors during state updates required strict input state clearing (setQuery('')) and locale enforcement.
Local Healthcare Data Structure: Structuring multi-variable filters (exam types, specific area bounds like Mirpur or Dhanmondi, and doctor gender preferences) so that query responses remain instantaneous and highly accurate.
UI/UX Polish under Time Constraints: Maintaining high visual standards—such as integrating mandatory hackathon badges with precise aspect ratios (height: 36px, width: auto) without breaking responsive navigation bars.
📚 What We Learned Building HealthMatch-BD taught us how critical localized AI and intuitive UI design are when addressing regional challenges:
Localized UX Matters: Adapting speech recognition to local dialects (bn-BD) drastically lowers the barrier to entry for users who aren't comfortable typing complex medical terms in English.
Efficient Database Querying: Designing streamlined relational schemas in PostgreSQL allowed us to handle complex multi-criteria matching efficiently.
Component Optimization: We deepened our understanding of state lifecycle management in Next.js when combining real-time audio event listeners with dynamic user interfaces.
Built With
- ai
- javascript
- next.js
- node.js
- react
- tailwind
- typescript