Data-Centric-FM-Healthcare

KO | EN

gitlite — search

#react #ai #typescript #go #machine-learning #deep-learning #docker #macos #java #windows #cli #kubernetes #awesome-list #openai

Data-Centric-FM-Healthcare

(★ 78)

A survey on data-centric foundation models in healthcare.

Data-Centric-FM-Healthcare 최신버젼 다운로드

최종 버전 다운로드 (.zip)

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

# Data-Centric Foundation Models in Computational Healthcare

:fire::fire::fire: **A survey on data-centric foundation models in computational healthcare**

**[Project Page](https://data-centric-fm-healthcare.github.io/)** | **[Paper [arXiv]](https://arxiv.org/abs/2401.02458)**

Last updated: 2026/02/05

:pencil: **If you find this repo helps, please kindly cite our survey, thanks!**

```bibtex
@article{zhang2024data,
  title={Data-Centric Foundation Models in Computational Healthcare: A Survey},
  author={Zhang, Yunkun and Gao, Jin and Tan, Zheling and Zhou, Lingfeng and Ding, Kexin and Zhou, Mu and Zhang, Shaoting and Wang, Dequan},
  journal={arXiv},
  year={2024},
  eprint={2401.02458},
  archivePrefix={arXiv},
  primaryClass={cs.LG},
  doi={10.48550/arXiv.2401.02458},
  url={https://arxiv.org/abs/2401.02458}
}
```

---

In this repository, we provide an up-to-date list of healthcare-related foundation models and datasets, which are also mentioned in our survey paper.

:book: **Contents**

- [Healthcare and medical foundation models](#Healthcare-and-medical-foundation-models)
  - [Language models](#Language-Models)
  - [Vision models](#Vision-Models)
  - [Vision-language models](#Vision-Language-Models)
  - [Protein and molecule models](#Protein-and-Molecule-Models)
  - [Other models](#Other-Models)
- [Datasets for foundation model](#Datasets-for-foundation-model)
  - [Text](#Text)
  - [Imaging](#Imaging)
  - [Genomics](#Genomics)
  - [Drug](#Drug)
  - [Multi-modal](#Multi-Modal)

---

## Healthcare and Medical Foundation Models

A star (*) after the pre-training data shows that the authors constructed the data with more than three sources.

### Language Models

### Vision Models

### Vision-Language Models

### Protein and Molecule Models

### Other Models

| Model     | Subfield            | Paper                                                                                                                                | Code | Base        | Pre-Training Data |
|:--------- |:------------------- |:------------------------------------------------------------------------------------------------------------------------------------ |:---- |:----------- |:----------------- |
| OmniNA    | Nucleotide sequence | [OmniNA: A Foundation Model for Nucleotide Sequences](https://www.biorxiv.org/content/10.1101/2024.01.14.575543.abstract)            | -    | LLaMA       | NCBI              |
| LaBraM    | EEG                 | [Large Brain Model for Learning Generic Representations with Tremendous EEG Data in BCI](https://openreview.net/forum?id=QzTpTRVtrP) | -    | Transformer | *                 |
| Neuro-GPT | EEG                 | [Neuro-GPT: Towards A Foundation Model For EEG](https://doi.org/10.1109/ISBI56570.2024.10635453)                                                 | -    | -           | TUH EEG           |

## Datasets for Foundation Model

### Text

| Dataset (Paper)                                              | Description                                                  | Link                                                         |
| :----------------------------------------------------------- | :----------------------------------------------------------- | :----------------------------------------------------------- |
| MedBench ([DOI](https://doi.org/10.26599/bdma.2024.9020044))         | A Chinese medical LLM benchmark with 300,901 Chinese questions covering 43 clinical specialties, combined with an automatic evaluation system | [Official site](https://medbench.opencompass.org.cn)         |
| MMedBench ([DOI](https://doi.org/10.1038/s41467-024-52417-z))        | A multilingual medical QA benchmark, where questions are categorized into 21 topics | [Github](https://github.com/MAGIC-AI4Med/MMedLM)             |
| MMedC ([DOI](https://doi.org/10.1038/s41467-024-52417-z))            | A multilingual medical corpus containing over 25.5B tokens   | [Github](https://github.com/MAGIC-AI4Med/MMedLM)             |
| BiMed1.3M ([DOI](https://doi.org/10.18653/v1/2024.findings-emnlp.989))        | An English and Arabic bilingual dataset of 1.3M samples of medical QA and chat | [Github](https://github.com/mbzuai-oryx/BiMediX)             |
| GAP-Replay ([arXiv](https://arxiv.org/abs/2311.16079))       | 48.1B tokens from 4 medical corpora including guidelines, abstracts, papers, and replay | [Github](https://github.com/epfLLM/meditron)                 |
| Huatuo-26M ([DOI](https://doi.org/10.18653/v1/2025.findings-naacl.211))       | 26M Chinese medical QA pairs                                 | [Github](https://github.com/FreedomIntelligence/Huatuo-26M)  |
| Medical Meadow ([arXiv](https://arxiv.org/abs/2304.08247))   | 16M medical QA pairs collected from 9 sources                | [Github](https://github.com/kbressem/medAlpaca)              |
| MultiMedQA ([Nature](https://www.nature.com/articles/s41586-023-06291-2)) | 6 existing and 1 online-collected medical QA dataset         | [Nature](https://www.nature.com/articles/s41586-023-06291-2#data-availability) |
| BigBio ([NeurIPS](https://proceedings.neurips.cc/paper_files/paper/2022/hash/a583d2197eafc4afdd41f5b8765555c5-Abstract-Datasets_and_Benchmarks.html)) | 126+ biomedical NLP datasets covering 13 task categories and 10+ languages | [Github](https://github.com/bigscience-workshop/biomedical)  |
| MedMCQA ([MLR](https://proceedings.mlr.press/v174/pal22a.html)) | 194K multiple-choice questions covering 2.4K healthcare topics | [Official site](https://medmcqa.github.io/)                  |
| MedQA-USMLE ([MDPI](https://www.mdpi.com/2076-3417/11/14/6421)) | 61,097 multiple choice questions based on USMLE in three languages | [Github](https://github.com/jind11/MedQA)                    |
| CBLUE ([DOI](https://doi.org/10.18653/v1/2022.acl-long.544))            | A Chinese biomedical language understanding evaluation benchmark with 18 datasets | [Official site](https://tianchi.aliyun.com/dataset/95414)    |
| BLURB ([DOI](https://doi.org/10.1145/3458754))            | 13 biomedical NLP datasets in 6 tasks                        | [Official site](https://microsoft.github.io/BLURB/index.html) |
| PubMedQA ([DOI](https://doi.org/10.18653/v1/d19-1259))         | 1K expert-annotated, 61.2K unlabeled, and 211.3K artificially generated biomedical QA instances | [Official site](https://pubmedqa.github.io/)                 |
| BLUE ([DOI](https://doi.org/10.18653/v1/w19-5006))             | 5 language tasks with 10 biomedical and clinical text datasets | [Github](https://github.com/ncbi-nlp/BLUE_Benchmark)         |
| webMedQA ([BMC](https://bmcmedinformdecismak.biomedcentral.com/articles/10.1186/s12911-019-0761-8)) | 63,284 real-world Chinese medical questions with over 300K answers | [Github](https://github.com/hejunqing/webMedQA)              |
| MedMentions ([DOI](https://doi.org/10.24432/C5G59C))      | 4,392 papers annotated by experts with mentions of UMLS entities | [Github](https://github.com/chanzuckerberg/MedMentions)      |
| MIMIC-III ([Nature](https://www.nature.com/articles/sdata201635)) | Critical care data for over 40,000 patients                  | [Official site](https://mimic.mit.edu/docs/iii/)             |
| ClinicalTrials.gov                                           | An online database of clinical research studies, including clinical trials and observational studies | [Official site](https://clinicaltrials.gov)                  |

### Imaging

| Dataset (Paper)                                              | Description                                                  | Link                                                         |
| :----------------------------------------------------------- | :----------------------------------------------------------- | :----------------------------------------------------------- |
| 3M-US                                                        | 2,187,915 ultrasound images of 12 common organs              | -                                                            |
| AbdomenAtlas ([arXiv](https://arxiv.org/abs/2407.16697))     | 20,460 3D CT volumes from 112 hospitals, with 673K masks of anatomical structures | [Github](https://github.com/MrGiovanni/AbdomenAtlas)         |
| BiomedParseData ([Nature](https://www.nature.com/articles/s41592-024-02499-w)) | 1.1M images, 3.4M image-mask-label triples, and 6.8M image-mask-description triples | [Github](https://github.com/microsoft/BiomedParse)           |
| Mass-100K ([DOI](https://doi.org/10.1038/s41591-024-02857-3))        | 100M tissue patches from 100,426 diagnostic H&E WSIs across 20 major tissue types | -                                                            |
| RETFound ([Nature](https://www.nature.com/articles/s41586-023-06555-x)) | Unannotated retinal images, containing 904,170 CFPs and 736,442 OCT scans | [Nature](https://www.nature.com/articles/s41586-023-06555-x#data-availability) |
| AbdomenAtlas-8K ([NeurIPS](https://proceedings.neurips.cc/paper_files/paper/2023/hash/7331077e0449e94a91370c46b4f80f57-Abstract-Datasets_and_Benchmarks.html))  | 8,448 CT volumes with per-voxel annotated eight abdominal organs | [Github](https://github.com/MrGiovanni/AbdomenAtlas)         |
| Med-MNIST v2 ([Nature](https://www.nature.com/articles/s41597-022-01721-8)) | 12 2D and 6 3D datasets for biomedical image classification  | [Official site](https://medmnist.com/)                       |
| EchoNet-Dynamic ([DOI](https://doi.org/10.1038/s41586-020-2145-8)) | 10,030 expert-annotated echocardiogram videos                | [Official site](https://echonet.github.io/dynamic/)          |
| CheXpert ([DOI](https://doi.org/10.1609/AAAI.V33I01.3301590))         | 224,316 chest radiographs of 65,240 patients                 | [Official site](https://stanfordmlgroup.github.io/competitions/chexpert/) |
| Kather Colon Dataset ([PMC](https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6345440/)) | 100K histological images of human colorectal cancer and healthy tissue | [Zenodo](https://zenodo.org/records/1214456)                 |
| DeepLesion ([PMC](https://pubmed.ncbi.nlm.nih.gov/30035154/)) | 32K CT scans with annotations and semantic labels from radiological reports | [NIH](https://www.nih.gov/news-events/news-releases/nih-clinical-center-releases-dataset-32000-ct-images) |
| ChestXray-NIHCC ([DOI](https://doi.org/10.1109/CVPR.2017.369))  | 100K radiographs with labels from more than 30,000 patients  | [NIH](https://nihcc.app.box.com/v/ChestXray-NIHCC)           |
| ISIC                                                         | An archive containing 23K skin lesion images with labels & Imaging | [Official site](https://www.isic-archive.com/)               |

### Genomics

| Dataset (Paper)                                                              | Description                                                                                                   | Link                                                         |
|:---------------------------------------------------------------------------- |:------------------------------------------------------------------------------------------------------------- |:------------------------------------------------------------ |
| 1000 Genomes Project ([Nature](https://www.nature.com/articles/nature15393)) | A comprehensive catalog of human genetic variations                                                           | [Official site](https://www.internationalgenome.org/)        |
| ENCODE ([Nature](https://www.nature.com/articles/nature11247))               | A platform of genomics data and encyclopedia with integrative-level and ground-level annotations              | [NIH](https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3439153/) |
| dbSNP ([NIH](https://www.ncbi.nlm.nih.gov/pmc/articles/PMC29783/))           | A collection of human single nucleotide variations, microsatellites, and small-scale insertions and deletions | [NIH](https://pubmed.ncbi.nlm.nih.gov/11125122/)             |

### Drug

| Dataset (Paper)                                                         | Description                                                                                     | Link                                            |
|:----------------------------------------------------------------------- |:----------------------------------------------------------------------------------------------- |:----------------------------------------------- |
| DrugChat ([arXiv](https://arxiv.org/abs/2309.03907))                    | 143,517 question-answer pairs covering 10,834 drug compounds, collected from PubChem and ChEMBL | - |
| PubChem ([NIH](https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9825602/))  | A collection of 900+ sources of chemical information data                                       | [NIH](https://pubchem.ncbi.nlm.nih.gov/)        |
| DrugBank ([NIH](https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5753335/)) | A web-enabled structured database of molecular information about drugs                          | [Official site](https://www.drugbank.com/)      |
| ChEMBL ([NIH](https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3245175/))   | 20M bioactivity measurements for 2.4M distinct compounds and 15K protein targets                | [Official site](https://www.ebi.ac.uk/chembl/)  |

### Multi-Modal

| Dataset (Paper)                                              | Description                                                  | Link                                                         |
| :----------------------------------------------------------- | :----------------------------------------------------------- | :----------------------------------------------------------- |
| MultiMedBench ([NEJM AI](https://doi.org/10.1056/AIoa2300138)) | A multi-modal benchmark comprising 12 data sources and 14 tasks | -                                                            |
| RadGenome-Chest CT ([arXiv](https://arxiv.org/abs/2404.16754)) | A dataset of 3D chest CT, including 197 organ-level segmentation masks, 665K multi-granularity grounded reports, and 1.3M grounded VQA pairs | -                                                            |
| OmniMedVQA ([DOI](https://doi.org/10.1109/CVPR52733.2024.02093))       | 131,813 question-answering items with 120,530 images from 12 modalities and 26 human anatomical regions, collected from 75 medical datasets | -                                                            |
| SAT-DS ([DOI](https://doi.org/10.1038/s41746-025-01964-w))           | 11,462 scans with 142,254 segmentation annotations spanning 8 human body regions from 31 medical image segmentation datasets, together with domain knowledge from e-Anatomy and UMLS | [Github](https://github.com/zhaoziheng/SAT)                  |
| PathChatInstruct ([DOI](https://doi.org/10.1038/s41587-024-02326-9)) | 250K+ diverse disease-agnostic visual-language instructions with image and text | -                                                            |
| Chi-Med-VL ([arXiv](https://arxiv.org/abs/2310.17956))       | 580,014 image-text pairs and 469,441 question-answer pairs for general healthcare in Chinese | [Github](https://github.com/williamliujl/Qilin-Med-VL)       |
| MedMD ([DOI](https://doi.org/10.1038/s41467-025-62385-7))            | 15.5M 2D scans and 180k 3D radiology scans  with textual descriptions | [Github](https://github.com/chaoyi-wu/RadFM)                 |
| OpenPath ([Nature](https://www.nature.com/articles/s41591-023-02504-3)) | 208,414 pathology images paired with natural language descriptions | [Huggingface](https://huggingface.co/spaces/vinid/webplip)   |
| Quilt-1M ([NeurIPS](https://proceedings.neurips.cc/paper_files/paper/2023/hash/775ec578876fa6812c062644964b9870-Abstract-Datasets_and_Benchmarks.html))         | 1M image-text pairs for histopathology                       | [Github](https://github.com/wisdomikezogwo/quilt1m)          |
| Med-MMHL ([arXiv](https://arxiv.org/abs/2306.08871))         | Human- and LLM-generated misinformation detection dataset    | [Github](https://github.com/styxsys0927/Med-MMHL)            |
| Mol-Instructions ([OpenReview](https://openreview.net/forum?id=Tlsdsb6l9n)) | 148K molecule-oriented, 505K protein-oriented, and biomolecular text instructions | [Huggingface](https://huggingface.co/datasets/zjunlp/Mol-Instructions) |
| PathInstruct ([DOI](https://doi.org/10.1609/AAAI.V38I5.28308))     | 180K samples of LLM-generated instruction-following data     | [Github](https://github.com/superjamessyx/Generative-Foundation-AI-Assistant-for-Pathology) |
| PMC-VQA ([arXiv](https://arxiv.org/abs/2305.10415))          | 227K VQA pairs of 149K images of various modalities or diseases | [Github](https://github.com/xiaoman-zhang/PMC-VQA)           |
| PMC-OA ([DOI](https://doi.org/10.1007/978-3-031-43993-3_51))           | 1.6M fine-grained biomedical image-text pairs                | [Github](https://github.com/WeixiongLin/PMC-CLIP)            |
| PathCap ([DOI](https://doi.org/10.1007/978-3-031-43993-3_51))          | 142K pathology image-caption pairs from various sources      | [Github](https://github.com/WeixiongLin/PMC-CLIP)            |
| SwissProtCLAP ([DOI](https://doi.org/10.1038/s42256-025-01011-z))    | 441K text-protein sequence pairs                             | [Github](https://github.com/chao1224/chatdrug)               |
| MIMIC-IV ([Nature](https://www.nature.com/articles/s41597-022-01899-x)) | Clinical information for hospital stays of over 60,000 patients | [Official site](https://mimic.mit.edu/docs/iv/)              |
| MIMIC-CXR ([Nature](https://www.nature.com/articles/s41597-019-0322-0)) | 227,835 chest imaging studies with free-text reports for 65,379 patients | [PhysioNet](https://physionet.org/content/mimic-cxr/2.0.0/)  |
| TCGA                                                         | A landmark cancer genomics program, molecularly characterized over 20,000 primary cancer and matched normal samples spanning 33 cancer types | [Official site](https://www.cancer.gov/ccg/research/genome-sequencing/tcga) |

Data-Centric-FM-Healthcare

Data-Centric-FM-Healthcare 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

Data-Centric-FM-Healthcare

Data-Centric-FM-Healthcare 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

Download README