README
Synthetic dataset:
generated.csv - synthetic datasets containing 41,185 clinical note samples spanning 219 ICD-10 codes.
| Data field | Description |
|---|---|
| idx | Unique sample identifier. |
| ICD-10 | The targeted ICD-10 code used for prior data sampling. |
| generation_model | The model used for sample generation (GTP-3.5, GPT-4, LLaMA-7b, LLaMA-13b) |
| prompt | Prompt used for sample generation. |
| prior | Type of prior data used for sample generation. |
| example | Bool variable for the presence or absence of example during generation. |
| example | source Source of example (open-source RuMedPrime or private medical data). |
| response | Result of model generation. |
| symptoms | Symptoms used for prompt creation. |
| anamnesis | Clinical note example used as a style example in the prompt. |
| symptoms_recall | BERT-score for response and symptoms. |
| anamnesis_precision | BERT-score for response and anamnesis |
Part of real in-house clinical notes was hidden and replaced with private_data mark.
Thirty samples from private real data were completely anonymized (manually by humans) and preserved in the dataset.
Citation
@InProceedings{10.1007/978-3-031-70381-2_14,
author="Kumichev, Gleb
and Blinov, Pavel
and Kuzkina, Yulia
and Goncharov, Vasily
and Zubkova, Galina
and Zenovkin, Nikolai
and Goncharov, Aleksei
and Savchenko, Andrey",
editor="Bifet, Albert
and Krilavi{\v{c}}ius, Tomas
and Miliou, Ioanna
and Nowaczyk, Slawomir",
title="MedSyn: LLM-Based Synthetic Medical Text Generation Framework",
booktitle="Machine Learning and Knowledge Discovery in Databases. Applied Data Science Track",
year="2024",
publisher="Springer Nature Switzerland",
address="Cham",
pages="215--230",
isbn="978-3-031-70381-2"
}
README
Synthetic dataset:
generated.csv - synthetic datasets containing 41,185 clinical note samples spanning 219 ICD-10 codes.
| Data field | Description |
|---|---|
| idx | Unique sample identifier. |
| ICD-10 | The targeted ICD-10 code used for prior data sampling. |
| generation_model | The model used for sample generation (GTP-3.5, GPT-4, LLaMA-7b, LLaMA-13b) |
| prompt | Prompt used for sample generation. |
| prior | Type of prior data used for sample generation. |
| example | Bool variable for the presence or absence of example during generation. |
| example | source Source of example (open-source RuMedPrime or private medical data). |
| response | Result of model generation. |
| symptoms | Symptoms used for prompt creation. |
| anamnesis | Clinical note example used as a style example in the prompt. |
| symptoms_recall | BERT-score for response and symptoms. |
| anamnesis_precision | BERT-score for response and anamnesis |
Part of real in-house clinical notes was hidden and replaced with private_data mark.
Thirty samples from private real data were completely anonymized (manually by humans) and preserved in the dataset.
Citation
@InProceedings{10.1007/978-3-031-70381-2_14,
author="Kumichev, Gleb
and Blinov, Pavel
and Kuzkina, Yulia
and Goncharov, Vasily
and Zubkova, Galina
and Zenovkin, Nikolai
and Goncharov, Aleksei
and Savchenko, Andrey",
editor="Bifet, Albert
and Krilavi{\v{c}}ius, Tomas
and Miliou, Ioanna
and Nowaczyk, Slawomir",
title="MedSyn: LLM-Based Synthetic Medical Text Generation Framework",
booktitle="Machine Learning and Knowledge Discovery in Databases. Applied Data Science Track",
year="2024",
publisher="Springer Nature Switzerland",
address="Cham",
pages="215--230",
isbn="978-3-031-70381-2"
}