--- language: - uk license: apache-2.0 tags: - ukrainian - medical - healthcare - clinical-protocols - instruction-tuning - rag - qa - public-health - telemedicine - clinical-decision-support - low-resource - ukrainian-nlp --- # ⚠️ LIMITED SAMPLE ONLY. GET THE FULL ACCESS. ## ⚕️ UkrMedProtocols-1 — Ukrainian Medical Guidelines Dataset (Demo) > A structured Q\&A dataset extracted from official public clinical protocols published by the Ministry of Health of Ukraine, designed for clinical AI and RAG applications. This repository contains a **demo version** (50 high-quality instruction-following QA pairs) from the full **UkrMedProtocols-1** dataset (2,512 QA pairs). ## 📢 WHY YOU SHOULD BUY THE FULL VERSION (2,512 QA Pairs) **This is a small, free preview. The full dataset is significantly more comprehensive and essential for creating reliable clinical decision support systems in Ukrainian.** ### ✨ Key Benefits of the Full UkrMedProtocols-1 Edition * **Official Ukrainian Medical Knowledge:** Verbatim content extracted from a real MOH clinical protocol, structured specifically for AI consumption. * **Precise, Context-Rich Answers:** Includes specific dosage guidelines, safety procedures, antibiotic timing rules, patient categorization, and diagnostic criteria. * **Field-Medicine Relevant Focus:** Covers critical areas like infection prevention in trauma, wound management, and standard safety precautions, making it highly practical. * **Ready for Fine-Tuning & RAG:** Clean Alpaca format, compatible with Axolotl, Unsloth, Llama-Factory, and specialized clinical AI pipelines. ## 🔑 GET THE FULL ENTERPRISE EDITION NOW! * Full Commercial/Research License * Full JSONL structure (2,512 high-quality QA pairs) * Reliable, verified clinical information **€50** https://www.patreon.com/posts/145552473 **Shoot me an email to arrange a licensing agreement:** founder@davidlab.tech --- ## 📜 License (This Demo) **MIT License.** You can freely use, modify, and fine-tune models on this **demo** under the MIT License. ***The full UkrMedProtocols-1 dataset (2,512 pairs) is not open-source and is distributed under a separate commercial/research license.*** *Note: The source content is public domain as an official document of the Ministry of Health of Ukraine.*