My Products
Projects I've Led and Developed
🏥 H. Pylori Research Data Center
A comprehensive cloud-native data platform for Helicobacter pylori (H. pylori) research, integrating global strain data, virulence factors, antibiotic resistance, and gastrointestinal disease records. The platform provides real-time interactive heatmaps, advanced search, data analysis dashboards, and AI-assisted medical imaging segmentation.
🗂 System Architecture
Full-stack architecture covering data ingestion pipelines, RESTful API layer, PostgreSQL with materialized views for performance, and an interactive frontend dashboard.
🧹 Automated Data Cleaning Pipeline
Built an end-to-end automated pipeline to ingest raw genomic and clinical data from multiple sources, clean and normalize records, resolve duplicates, and load into PostgreSQL — reducing manual processing time by over 90%.
🌍 Global Interactive Heatmap
Real-time interactive world heatmaps covering H. pylori strain virulence factors, antibiotic resistance, gastrointestinal diseases, and patient distribution — all filterable by date range and region.
🔍 Advanced Search Module
Multi-dimensional search across strain records, supporting filtering by country, region, antibiotic resistance profile, virulence factors, disease type, and date — with paginated results and CSV export.
📊 Data Analysis Dashboard
Interactive analytics with Chart.js visualizations — time-series trends, country/region distribution, disease co-occurrence, resistance pattern heatmaps, and statistical summaries powered by R and Python pipelines.
🔬 AI Medical Imaging Segmentation Module
Integrated an AI-powered medical imaging segmentation module into the platform, enabling automated analysis of endoscopic images and visualization of lesion areas to assist clinical research.
🤖 AI Agent Platform & Federated Learning System
A multi-product AI platform combining an intelligent medical agent system and a federated learning framework. The platform includes a hospital monitoring & control console, a patient-facing AI assistant, and a distributed federated learning pipeline that enables privacy-preserving model training across multiple hospital nodes without sharing raw patient data.
🗂 System Architecture
Overall product architecture covering the hospital admin console, patient-facing AI agent, federated learning coordination layer, and shared data infrastructure.
🏥 Hospital Monitoring & Control Console
A real-time hospital management dashboard for administrators — monitoring patient flows, device status, ward occupancy, and operational metrics. Supports multi-site deployment across hospital branches with role-based access control.
💬 Patient-Facing AI Assistant
An intelligent conversational AI assistant for patients, powered by a locally deployed Llama model with RAG (Retrieval-Augmented Generation) over a medical knowledge base. Patients can query symptoms, medications, and appointment information in natural language.
🔗 https://562lmj.zeabur.app/🔗 Federated Learning Framework
A privacy-preserving federated learning pipeline that enables multiple hospital nodes to collaboratively train AI models without sharing raw patient data. Each node trains locally and only shares model gradients with a central aggregation server.