DocMe Assistant
Intelligent chatbot to improve access to medical services in Colombia
- Client
- Academic project · DocMe
- Role
- Backend Developer · Machine Learning
- Year
- 2024
Summary — I designed and built the full chatbot architecture, from TensorFlow/Keras model training to Flask integration with MySQL for users and appointments.
Colombia's healthcare system faces overloaded hotlines and delays accessing medical services. DocMe addresses these limitations with a conversational assistant that classifies intents, schedules appointments, and provides basic emotional support.
NLP + neural networks
I trained neural networks with TensorFlow and Keras to classify user intents. Combined with NLTK and PySpellChecker for natural language processing and spell correction, the bot understands greetings, symptoms, emotional support requests, and appointment requests.
- Models
- TensorFlow · Keras
- NLP
- NLTK · PySpellChecker
What I contributed
- End-to-end chatbot architecture.
- Intent classification model training.
- Flask backend with SQLAlchemy + MySQL.
- Production server with Waitress (WSGI).
- NLP processing, spell correction, and responses.