SB
Tech.
Clinical Intelligence

AURA HomeoAI

Clinical Decision Support & Agentic Case Repertorization Suite

A state-of-the-art clinical helper for homeopathic practitioners. Incorporating anamnesis dictation, document extracting, computer vision symptom parsing (skin lesions, rashes), and an autonomous web-search loop that queries remedies and Materia Medica modals while maintaining local confidentiality.

Explore Repository View Flow
Patient Anamnesis Dashboard
Patient ID #PT-2026-09
Age / Gender 34 / Female
Chief Complaint (Symptom 1 - Priority)
"Eczema on right forearm, intensely itchy, worse from warmth..."
Diagnostic Upload
lesion_forearm.jpg Analyzed via local llava: skin rash with erythema
Agent Web-Search repertorization loop... Cycle 3/8

> DDG Query: "eczema worse warmth homeopathic remedy"

> Synthesizing Modal Totality: Sulphur (94% Match), Mezereum (78% Match)

Stateful ReAct

8-Cycle Synthesis Loop

Vision Parser

Auto-Discovery Ollama Tags

Voice Dictation

Native Web Speech API

Zero-Trust

Neon Cloud License Auth

Diagnostic Workflow Flowchart

AuraHomeoAI.exe 5MB Compiled Launcher Neon Postgres Cloud License Auth Flask app.py Web Dashboard SQLite cases.db Local Cache history Agent reasoning Loop ReAct (Materia Medica) Multi-modal inputs PDFs, Images, Web Speech DDG Dual-Path Scraper Library & Regex Fallback

Advanced Engineering Core

ReAct Totality Analysis

Rather than executing one-shot analysis, AURA HomeoAI implements the ReAct loop framework. It iterates up to 8 reasoning and research cycles, comparing symptoms against Materia Medica resources before exporting a unified JSON database model.

Multi-Modal Vision Pipeline

Automatically scans local Ollama system tag catalogs at boot to discover vision-capable models (like `llava` or `qwen2-vl`). Renders skin rash uploads or medical imaging directly into objective, text-based symptom prompts.

Dual-Path Web Scraping

Ensures maximum uptime. Utilizes DuckDuckGo Search libraries as its primary search channel, falling back automatically to a custom raw HTTP parser using regex expressions if API connections drop.

Zero-Trust Licensing Hook

Before starting the server, AURA HomeoAI routes licensing validations through a remote, secure cloud Neon PostgreSQL database. Gracefully resolves local networking dropouts using offline fallbacks.

Tech Stack & Infrastructure

Logic & Database

  • Python 3.10+
  • Flask Web Framework
  • SQLite3 Cache
  • Neon Cloud Postgres

AI Models

  • Ollama local host
  • Qwen 3 (8B)
  • Llama 3.1 (8B)
  • Llava / Llama Vision

Frontend & Compilation

  • Vanilla Javascript (ES6+)
  • Web Speech Voice Dictation
  • Tailwind / Glassmorphism CSS
  • PyInstaller executable build

Execution & Startup

AURA HomeoAI is compiled as a lightweight 5MB launcher executable (`AuraHomeoAI.exe`), abstracting runtime packaging. Launching the EXE triggers an autostart pipeline:

Step 1

Checks authentication credentials against cloud Neon PostgreSQL.

Step 2

Autostarts local Ollama service in background and pulls Qwen3 if missing.

Step 3

Spins up Flask server on port 5000 and opens default browser dashboard.