HEIDI: Multilingual AI Health Chatbot to Support Elderly, Caregivers and Migrant Communities

Tags: #Healthcare #MultiLingual #GenerativeAI #LLM #Chatbot

Background

In Singapore's multilingual society, where health literacy directly impacts healthcare outcomes, Analytico AI developed HEIDI (Health Empowerment via Informed Decisions & Intelligence) to address a critical challenge to improve healthcare access and literacy for elderly residents, foreign domestic workers, and migrant communities in Singapore. This innovative AI chatbot solution redefines healthcare communication by providing accessible, reliable medical information to Singapore residents aged 18-80, supporting the nation's healthcare goals and empowering users to take charge of their well-being.

HEIDI leverages cutting-edge AI technology to transform how individuals access and understand health information. The platform combines advanced language models with curated medical databases to deliver trustworthy health insights that help users make informed healthcare decisions.

Key Features

• Intelligent Medical Information Retrieval: Engage with specialized medical databases including Cleveland Clinic, HealthHub, and Ministry of Health resources. The system provides condition-specific information with symptoms, treatments, and preventive measures through an intuitive dropdown selection interface.

• Natural Language Processing with Empathy: Powered by advanced LLM models (Gemini Flash), the chatbot understands context and provides human-like responses with appropriate emotional sensitivity, including query clarification for accurate information delivery.

• Multi-Modal & Multilingual Support: Accept queries through text, voice, or image inputs, with responses available in text and audio. Full support for English, Chinese, Malay, Tamil, Bengali, Thai, Tagalog and Singlish, ensuring accessibility across Singapore's diverse population.

Technology & Innovation

The system incorporates three specialised components to enhance healthcare accessibility:

Retrieval-Augmented Generation (RAG) Architecture:

A sophisticated system combining embedding models, vector database (FAISS), and large language models to convert medical content into manageable chunks for efficient retrieval and response generation. This ensures users receive accurate, contextually relevant information with consistent response times under 3 seconds.

Comprehensive Analytics Dashboard:

Real-time tracking of user engagement metrics, query patterns, and source attribution. Features include word clouds for common health topics, retention analysis, and detailed usage statistics to provide valuable insights into community health information needs.

AI Safety & Reliability Framework:

Regular evaluation using Project Moonshot, content filtering for harmful advice, clear emergency disclaimers, and continuous model retraining. The system maintains 99.95% uptime with cloud-based auto-scaling infrastructure supporting up to 10,000+ concurrent users.

Impact on Healthcare Communication

By creating an inclusive health literacy platform that combines trustworthy medical information with culturally sensitive AI communication, HEIDI addresses fundamental challenges in Singapore's healthcare system. The chatbot not only improves access to health information but also prepares users for more productive healthcare professional interactions. With projected outcomes including a 25% increase in customer retention and 85% sourcing accuracy from TGLH-approved databases, this innovation represents a significant advancement in democratizing healthcare knowledge and empowering Singapore's diverse population to make informed health decisions.