AI Insights Chatbot (Turn feedback to insights) — NRC x FPT Hackathon 2025

NRC x FPT Hackathon 2025 — First Prize Winner

"The future of healthcare analytics lies not in data collection, but in intelligent understanding — where every patient voice becomes a pathway to actionable insights and better care." — NRC x FPT Hackathon 2025

---

🏆 Overview

Project: AI Insights Chatbot — Healthcare Intelligence Platform that Turns Feedback into Insights

Achievement: 🥇 First Prize, NRC x FPT Hackathon 2025

Event Theme: AI-Powered Healthcare Analytics

Date: January 2025

Team: Led by Kato (Quan Ngo)

---

🎯 Problem Statement

Healthcare facilities receive thousands of patient feedback comments daily — rich, unstructured text containing critical insights about services, conditions, and experiences. Yet, extracting meaningful and actionable insights from this feedback remains a monumental challenge.

Traditional analytics approaches fail to:

• Transform feedback into structured insights across different comments
• Generate actionable intelligence from sentiment context for specific healthcare entities
• Enable insight-driven semantic queries that mirror how healthcare professionals think
• Resolve entity variations (e.g., "heart condition" vs "cardiac issue") into unified insights

The challenge for NRC x FPT Hackathon 2025:

Build an intelligent chatbot system that transforms raw healthcare feedback into a structured knowledge graph of insights — enabling deep analysis, sentiment-aware entity tracking, and adaptive multi-agent reasoning to turn feedback into actionable insights.

---

🚀 Our Solution — AI Insights Chatbot with Multi-Agent Knowledge Graph RAG

Our winning solution combines three revolutionary technologies into a unified feedback-to-insights intelligence platform:

1. Knowledge Graph Ingestion Pipeline — Transforms unstructured feedback into semantically rich insights graph
2. Entity Extraction & Resolution — Identifies and unifies healthcare concepts to generate unified insights
3. Multi-Agent Adaptive RAG System — Intelligently routes queries through specialized agents to deliver actionable insights

Multi-Agent Application Architecture

---

🧠 Core Innovation: Feedback-to-Insights Knowledge Graph with Entity Resolution

Unique Entity Deduplication for Insight Generation

Unlike traditional systems that treat each mention separately, our pipeline ensures one unique entity node per normalized concept across all feedback, enabling consolidated insights:

• "Heart condition" and "cardiac issue" resolve to the same entity for unified insights
• Sentiment analysis aggregates at the entity level to generate comprehensive insights
• Healthcare professionals can query insights about concepts, not just keywords

Intelligent Entity Extraction for Insight Generation

Using AWS Bedrock Claude, we extract healthcare entities from feedback comments and transform them into actionable insights:

• Medical Conditions — Diseases, symptoms, diagnoses
• Healthcare Services — Treatments, procedures, consultations
• Facility Attributes — Staff quality, wait times, cleanliness
• Patient Experiences — Emotional states, satisfaction indicators

Each extracted entity is transformed into insights through:

• Normalization to a standard form for consistent insight generation
• Embedding with semantic vectors for insight similarity search
• Linking to feedback via sentiment-labeled relationships for insight context

Knowledge Graph RAG Architecture

---

💡 Sentiment-Aware Insight Generation

Contextual Sentiment Tracking for Deeper Insights

Our system doesn't just classify comments as positive or negative — it analyzes sentiment at the entity level to generate nuanced insights:

• A single feedback comment mentioning multiple entities gets individual insight scores for each
• Sentiment labels for insights include: positive, negative, and neutral
• Confidence scores accompany each insight assessment
• Original feedback spans are preserved for insight traceability

Rich Relationship Modeling for Connected Insights

The knowledge graph connects feedback to generate comprehensive insights:

• Patients → Provide feedback → Comments → Generate Insights
• Comments → About → Facilities → Create Facility Insights
• Comments → Relate to → Service Lines → Produce Service Insights
• Comments → Mention → Entities → Transform into Entity Insights (with sentiment properties)

This structure enables insight queries like:

• "What insights emerge about medical conditions mentioned with negative sentiment?"
• "Which facility insights show positive feedback about cardiac care?"
• "What insights reveal how patient experiences vary across different service lines?"

---

🤖 Multi-Agent Adaptive RAG System

Intelligent Query Routing

Our Coordinator Agent classifies incoming queries and routes them to specialized agents:

Semantic Search Agent

• Handles queries requiring similarity-based retrieval
• Leverages vector embeddings on both comments and entities
• Finds semantically related content even without exact keyword matches

Cypher Query Agent

• Executes graph traversal queries for relationship-based insights
• Navigates the knowledge graph structure efficiently
• Answers questions about connections and patterns

Reasoning Agent

• Performs complex analytical reasoning using Claude Sonnet
• Synthesizes information from multiple sources
• Generates insights that require multi-step logic

LangGraph Integration

The entire system is orchestrated using LangGraph, enabling intelligent insight generation:

• Dynamic agent selection based on insight complexity requirements
• Multi-step insight reasoning chains that combine different agent outputs
• Adaptive insight workflows that adjust based on intermediate results

---

📊 Feedback-to-Insights Data Flow Architecture

Ingestion Pipeline (Feedback → Insights)

1. Feedback Extraction — Pull structured healthcare feedback from SQL databases or CSV files
2. Entity Extraction for Insights — Use Bedrock Claude to identify healthcare concepts in feedback text
3. Sentiment Analysis for Insights — Analyze sentiment for each extracted entity to generate contextual insights
4. Embedding Generation for Insights — Create semantic vectors using Bedrock Titan embeddings for insight similarity
5. Insights Graph Construction — Build Neo4j knowledge graph with unique entity resolution for consolidated insights
6. Insights Index Creation — Set up property and vector indexes for fast insight retrieval

Query & Insights Analysis Flow

1. User Query → Coordinator Agent classifies the insight request type
2. Agent Selection → Routes to appropriate specialized insight-generation agent(s)
3. Insights Retrieval → Agents query the graph using semantic search or Cypher for relevant insights
4. Insight Reasoning & Synthesis → Combine results with multi-agent reasoning to generate actionable insights
5. Insights Response Generation → Deliver insights in natural language or structured formats

Export & Insights Reporting

• CSV Generation — Export insight query results for further analysis
• PDF Insight Reports — Generate formatted insight reports for stakeholders
• API Integration — Connect insights with existing healthcare systems

---

🎯 Key Highlights

• 🏆 First Prize Winner — Recognized for innovation in feedback-to-insights healthcare AI analytics
• 🔄 Feedback Transformation — Converts raw feedback into structured, actionable insights
• 🔗 Unique Entity Resolution — One entity per concept across all feedback sources for unified insights
• 💭 Sentiment-Aware Insights — Entity-level insight generation with confidence scores
• 🧠 Multi-Agent Intelligence — Adaptive routing for optimal insight generation and delivery
• 🔍 Semantic Insights Search — Vector embeddings enable natural language insight queries
• 📈 Scalable Insights Architecture — Designed to generate insights from millions of feedback comments and entities

First Prize Award — NRC x FPT Hackathon 2025

---

© 2025 Kato (Quan Ngo) — Team Lead, NRC x FPT Hackathon 2025 First Prize Winner