← Back to projects
AI-Powered Medical Text Summarization
PythonFlaskNLPTransformersKnowledge GraphsJavaScript
The Problem
Medical question-answer data is often lengthy and difficult for both patients and healthcare professionals to quickly interpret.
The Solution
Developed a hybrid deep learning architecture combining 8-head attention Transformers with BiLSTM and knowledge graphs to generate concise, context-aware summaries.
Architecture
RESTful Flask API serving a dual-summary generation pipeline with role-specific output formatting.
Role
AI/ML Engineer
Impact
Achieved 88%+ summarization accuracy with optimized inference time between 5–15 seconds under constrained hosting environments.
Lessons I Learned
Careful model optimization and attention tuning significantly improve contextual accuracy in domain-specific NLP tasks.