AI in Emergency Medicine - Turkiye (AI-EMT) is a community competition organized by Informatiqs. The goal is to predict two clinically critical outcomes for emergency department encounters across Turkiye using a large-scale tabular clinical dataset of 1,000,000 records spanning 2019 through 2024.

**Primary Target:** adverse_outcome - composite of ICU admission or in-hospital death

**Secondary Target:** readmission_30d - unplanned ED re-presentation within 30 calendar days

**Evaluation Metric:** Macro-Averaged AUC-ROC across both targets

**Dataset Size:** 1,000,000 encounters | 135 features | 25 cities across all NUTS-2 regions of Turkiye

---

## Competition Organizers

This competition is organized by the **Informatiqs** community.

| Role | Name |
|------|------|
| Community President | Kutluk Atalay |
| Community Vice President | Ozan Mohurcu |
| Lead Competition Organizer | Olaf Yunus Laitinen Imanov |

---

## Timeline

| Milestone | Date |
|-----------|------|
| Competition Start | Set via launch checklist |
| Team Merger Deadline | 5 days before close |
| Final Submission Deadline | See competition header |
| Private Leaderboard Release | Up to 7 days after close |

---

## Quick Start

Download the data, run the official starter notebook `setup_solution.ipynb`, and submit `submission.csv`. The starter notebook provides a working LightGBM baseline with 5-fold cross-validation achieving a macro AUC of approximately 0.85 on the training set.