Literature · OpenMed
OpenMed BloodCancerDetect
A compact 65M-parameter model that identifies chronic lymphocytic leukaemia clinical entities in biomedical text.
What OpenMed BloodCancerDetect is
OpenMed BloodCancerDetect (OpenMed-NER-BloodCancerDetect-TinyMed-65M) is a lightweight token-classification model developed by OpenMed. It is designed to recognise mentions of haematologic malignancies, with a dedicated focus on entities linked to chronic lymphocytic leukaemia (CLL), across medical records and scientific literature.
Extracting specialised clinical concepts, biomarkers, and therapies from unstructured biomedical writing is essential for oncology data processing. By isolating CLL-related entities, the model supports automated medical informatics workflows, literature curation, and clinical text mining.
How it works
The model employs a distilbert-base-cased architecture fine-tuned for token classification and named entity recognition. It takes tokenised biomedical text as input and applies the AdamW optimiser with learning rate scheduling during training, evaluated through cross-validation on a held-out test split.
Training is conducted on the domain-specific CLL corpus, an annotated dataset covering chronic lymphocytic leukaemia terminology, therapeutic interventions, and clinical concepts. The model outputs token labels using the BIO scheme, specifically identifying the beginning (`B-CL`) and continuation (`I-CL`) of leukaemia-related entities.
Key facts
- Base architecture
- distilbert-base-cased
- Parameter count
- 65M
- Task
- Token classification (named entity recognition)
- Target dataset
- CLL (Chronic Lymphocytic Leukemia) corpus
- Output labels
- B-CL, I-CL
What you can use it for
- Clinical Text Mining
- Extracting chronic lymphocytic leukaemia diagnoses, clinical concepts, and patient features from electronic health records.
- Biomedical Literature Analysis
- Parsing oncology and haematology journal articles to track mentions of leukaemia biomarkers and treatment entities.
- Knowledge Graph Construction
- Populating biomedical knowledge graphs with structured entities derived from haematological oncology research.
- Medication and Safety Monitoring
- Assisting in drug interaction detection and adverse event extraction within haematology documentation.
Run it here
No installation and no GPU of your own: fill in the inputs and it runs on our servers. Results stay in your browser unless you sign in.
Runs on On-demand GPU
Runs on our servers with on-demand compute. A first run needs time to load the model; active capacity can be reused and scales down when idle.Run this model
OpenMed BloodCancerDetect
Strengths and limits
Good at
- Achieves a high recall of 0.9434 on the CLL evaluation benchmark, capturing the vast majority of disease mentions.
- Compact 65-million-parameter DistilBERT base enables low-latency inference and manageable GPU memory requirements.
- Demonstrates an overall token classification accuracy of 0.9686 on domain-specific test text.
Watch out for
- Precision is comparatively lower than recall at 0.7812, resulting in occasional false-positive entity identifications.
- Prediction tags are limited exclusively to chronic lymphocytic leukaemia entities (`B-CL` and `I-CL`), offering no coverage for unrelated medical conditions.
- Achieves a lower overall F1 score (0.8547) than larger models in the same family, such as the 560M and 434M variants.
Inputs and outputs
You provide
Text
Entities are labelled token by token.
You get back
Labels with confidence scores, highest first.
Licence
Check the licence before commercial use · See model card
The See model card licence carries terms worth reading before you depend on it.
Questions
Can I use OpenMed BloodCancerDetect commercially?
Check the licence before commercial use. The See model card licence carries terms worth reading before you depend on it.
How do I run OpenMed BloodCancerDetect without installing anything?
Use the form under “Run it here” on this page: fill in the inputs and press Run model. It runs on our servers on CPU only, typically in 12s, and a run costs about 28 credits. Visitors without an account get 1,000 credits a month.
What entity types does this model extract?
It extracts chronic lymphocytic leukaemia entities tagged using the BIO scheme as either `B-CL` (beginning of entity) or `I-CL` (continuation of entity).
What architecture is used?
The model is based on distilbert-base-cased and fine-tuned for token classification.
What training data was used?
It was trained on the CLL corpus, a specialised biomedical dataset containing annotations for chronic lymphocytic leukaemia terminology, clinical concepts, biomarkers, and therapies.
How does this model compare to larger BloodCancerDetect models?
At 65 million parameters, it offers an F1 score of 0.8547, which is lower than the larger 560M ElectraMed model (0.9575 F1) but operates with fewer computational requirements.
How to cite it
Maziyar Panahi. OpenMed NER: Open-Source, Domain-Adapted State-of-the-Art Transformers for Biomedical NER Across 12 Public Datasets. arXiv:2508.01630 [cs.CL], 2025.
Written from OpenMed BloodCancerDetect's own documentation (model card). Statements citing a figure that does not appear there are removed before publication. Licence and running details come from our registry.