cardiomas 0.3.0.dev1

CardioMAS — Cardio Multi-Agent System for reproducible ECG dataset splits

CardioMAS — Cardio Multi-Agent System

A locally-runnable multi-agent system that analyzes ECG datasets, generates reproducible train/validation/test splits, and publishes split manifests (record IDs only) to vlbthambawita/ECGBench on HuggingFace.

Quick Start

pip install cardiomas

# Analyze a dataset and publish splits
cardiomas analyze https://physionet.org/content/ptb-xl/1.0.3/

# Check existing splits
cardiomas status ptb-xl

# List all analyzed datasets
cardiomas list --remote

Requirements

• Python ≥ 3.10
• Ollama running locally with a model pulled (default: llama3.1:8b)

ollama pull llama3.1:8b
ollama serve

Python API

from cardiomas import CardioMAS

mas = CardioMAS(ollama_model="llama3.1:8b", seed=42)
result = mas.analyze("https://physionet.org/content/ptb-xl/1.0.3/")

splits = mas.get_splits("ptb-xl")
train_ids = splits["train"]

Links

• HuggingFace Dataset: vlbthambawita/ECGBench
• PyPI: cardiomas
• GitHub: vlbthambawita/CardioMAS