op405wm/Bakalarska_praca
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
54b9229a40
Bakalarska_praca/README.md

907 B
Raw Blame History

DrugConv: Conversational Drug Recommendation System

DrugConv is a modular, Python-based backend + React frontend application that helps users find relevant medication recommendations, dosages and explanations via a friendly conversational interface powered by Mistral LLM and Elasticsearch.

Features

  • Conversational QA – Ask natural-language questions about medications, indications, dosing, and get follow-ups.
  • Mistral LLM integration – Dual-model (“small” & “large”) pipelines for generation, validation and evaluation.
  • Vector & text search – Hybrid retrieval from Elasticsearch for up-to-date drug documentation.
  • Memory & context – Tracks patient age, anamnesis, prescription type for multi-turn dialogue.
  • React SPA frontend – Built with React, styled with Tailwind CSS and Material UI for a clean, responsive UI.