Aremaki/LongBEL_8B_SPACCC

LongBEL: Long-Context and Document-Consistent Biomedical Entity Linking

LongBEL

LongBEL is a novel document-level framework for biomedical entity linking (BEL). Instead of normalizing each mention independently, LongBEL conditions each prediction on the document context and on previous normalizations produced in the same document. This design enforces document-level consistency and is enhanced by our robust memory mechanism. The method is introduced in our paper, currently under review.

LongBEL (SPACCC Edition)

This is a finetuned version of LLaMA-3-8B trained on SPACCC, applying the LongBEL framework to enable long context and robust memory predictions.

Field	Value
Base model	meta-llama/Llama-3.2-8B-Instruct
Task	Biomedical Entity Linking
Dataset	SPACCC
Knowledge base	SNOMED CT Spanish Version (July 31, 2021 release)
Input	BigBio-like documents with mention spans and semantic groups
Output	Ranked SNOMED concept predictions
Decoding	Semantic-guided constrained decoding
Main metric	Recall@1

Intended Use

This model is intended for research on biomedical entity linking and document-level consistency.

It assumes that mention spans and semantic groups are already provided. It does not perform named entity recognition. In a full pipeline, a NER model should first detect mentions and assign semantic groups, then LongBEL can normalize these mentions to SNOMED concepts.

Usage

Loading the model

import torch
from transformers import AutoModelForCausalLM

model = AutoModelForCausalLM.from_pretrained(
    "AnonymousARR42/LongBEL_8B_SPACCC",
    trust_remote_code=True,
    torch_dtype=torch.bfloat16,
    device_map="auto",
)

Inference example

The model expects BigBio-like documents. Each entity should include a mention text, character offsets, and a semantic group in the type field.

num_beams = 5

bigbio_pages = [
    {
        "id": "001",
        "document_id": "doc_001",
        "passages": [
            {
                "id": "0",
                "type": "paragraph",
                "text": [
                    "Una mujer embarazada de 29 años consultó por hipertensión grave, "
                    "cefalea y dolor epigástrico. Las pruebas de laboratorio mostraron proteinuria. "
                    "Fue ingresada durante la noche por sospecha de PET y se inició tratamiento urgente."
                ],
                "offsets": [[0, 227]],
            }
        ],
        "entities": [
            {
                "id": "T1",
                "type": "ENFERMEDAD",
                "text": ["hipertensión grave"],
                "offsets": [[45, 63]],
            },
            {
                "id": "T2",
                "type": "ENFERMEDAD",
                "text": ["proteinuria"],
                "offsets": [[131, 142]],
            },
            {
                "id": "T3",
                "type": "ENFERMEDAD",
                "text": ["PET"],
                "offsets": [[191, 194]],
            },
        ],
        "events": [],
        "coreferences": [],
        "relations": [],
    }
]

predictions = model.sample(
    bigbio_pages=bigbio_pages,
    num_beams=num_beams,
)

for i in range(0, len(predictions), num_beams):
    mention = predictions[i]["mention"]
    print(f"## Mention {(i // num_beams) + 1}: {mention}")

    for j in range(num_beams):
        pred = predictions[i + j]
        print(
            f"   - Beam {j + 1}:\n"
            f"     Predicted concept name: {pred['pred_concept_name']}\n"
            f"     Predicted code: {pred['pred_concept_code']}\n"
            f"     Beam score: {pred['beam_score']:.3f}\n"
        )

Example Output:

## Mention 1: hipertensión grave
   - Beam 1:
     Predicted concept name: hipertensión arterial
     Predicted code: 38341003
     Beam score: 0.993

   - Beam 2:
     Predicted concept name: degeneración vascular hipertensiva
     Predicted code: 38341003
     Beam score: 0.249

   - Beam 3:
     Predicted concept name: hipertensión arterial maligna
     Predicted code: 70272006
     Beam score: 0.046

   - Beam 4:
     Predicted concept name: degeneración macular senil
     Predicted code: 267718000
     Beam score: 0.004

   - Beam 5:
     Predicted concept name: hipertensión maligna secundaria, SAI
     Predicted code: 194784007
     Beam score: 0.001

## Mention 2: proteinuria
   - Beam 1:
     Predicted concept name: proteinuria de causa desconocida
     Predicted code: 231860006
     Beam score: 0.000

   - Beam 2:
     Predicted concept name: proteína de la membrana mitocondrial asociada con neurodegeneración
     Predicted code: 709415008
     Beam score: 0.000

   - Beam 3:
     Predicted concept name: proteinuria aislada concomitante con glomerulonefritis membranoproliferativa tipo III y debida a ella
     Predicted code: 368931000119104
     Beam score: 0.000

   - Beam 4:
     Predicted concept name: proteinosis alveolar pulmonar congénita
     Predicted code: 707442002
     Beam score: 0.000

   - Beam 5:
     Predicted concept name: proteinosis alveolar pulmonar
     Predicted code: 10501004
     Beam score: 0.000

## Mention 3: PET
   - Beam 1:
     Predicted concept name: preeclampsia
     Predicted code: 398254007
     Beam score: 0.285

   - Beam 2:
     Predicted concept name: preeclampsia en el puerperio
     Predicted code: 765182005
     Beam score: 0.068

   - Beam 3:
     Predicted concept name: púrpura trombocitopénica
     Predicted code: 302873008
     Beam score: 0.000

   - Beam 4:
     Predicted concept name: púrpura de la vulva
     Predicted code: 289487000
     Beam score: 0.000

   - Beam 5:
     Predicted concept name: pústula maligna
     Predicted code: 84980006
     Beam score: 0.000

Saliency map example

The model can also return token-level saliency maps during inference.

predictions, saliency_maps = model.sample(
    bigbio_pages=bigbio_pages,
    num_beams=num_beams,
    with_saliency_maps=True,
)

model.display_saliency_map(saliency_maps[2])

Example saliency map for the mention PET:

Evaluation

Entity linking performance is reported using Recall@1 with bootstrap confidence intervals. The best result is shown in bold, and the second-best result is underlined and ⭐ marks the main LongBEL-8B model.

Model	MM-ST21PV (English)	QUAERO-EMEA (French)	SympTEMIST (Spanish)	DisTEMIST (Spanish)	MedProcNER (Spanish)
Context-Free BEL
SciSpacy	53.8 ± 1.0	37.1 ± 4.3	9.8 ± 1.3	21.1 ± 1.9	10.3 ± 1.2
SapBERT	65.6 ± 1.0	59.7 ± 3.8	34.2 ± 2.0	38.6 ± 2.6	30.4 ± 2.1
CODER-all	62.9 ± 1.1	66.9 ± 4.0	42.2 ± 2.2	47.0 ± 2.6	42.7 ± 2.1
SapBERT-all	64.6 ± 1.1	67.9 ± 3.9	49.8 ± 2.4	49.6 ± 2.6	45.1 ± 2.2
BERGAMOT	60.9 ± 1.1	63.8 ± 4.9	48.0 ± 2.7	48.9 ± 2.4	42.3 ± 2.2
Local-Context BEL
ArboEL	76.9 ± 0.9	63.0 ± 3.9	55.4 ± 2.5	54.7 ± 2.6	59.7 ± 2.6
GENRE / mBART-large	69.6 ± 1.0	69.3 ± 5.4	59.8 ± 2.7	58.7 ± 2.7	66.0 ± 2.3
GENRE / Llama-1B	73.1 ± 1.0	75.1 ± 3.6	60.5 ± 2.4	62.5 ± 2.3	67.4 ± 2.1
GENRE / Llama-8B	75.0 ± 0.9	73.8 ± 4.0	61.7 ± 2.5	63.2 ± 2.5	68.3 ± 2.2
Global-Context BEL: LongBEL
LongBEL-1B	77.6 ± 0.9	74.5 ± 3.7	59.8 ± 2.5	61.9 ± 2.4	66.6 ± 2.1
LongBEL-1B + Ensemble	78.6 ± 0.8	77.2 ± 3.0	61.8 ± 2.5	64.3 ± 2.2	69.0 ± 2.0
⭐ LongBEL-8B	79.3 ± 0.8	75.4 ± 3.4	62.0 ± 2.6	63.6 ± 2.1	69.0 ± 2.1
LongBEL-8B + Ensemble	80.0 ± 0.8	77.6 ± 3.0	63.3 ± 2.5	65.8 ± 2.2	71.0 ± 2.0

The score reported for this checkpoint is the single LongBEL-8B model. The ensemble result requires fusing several LongBEL input configurations and is not produced by this checkpoint alone.

Speed and Memory

Measured on a single NVIDIA H100 80GB GPU.

Model	Model memory	Candidate memory	Speed
GENRE-Llama-8B baseline	28.6 GB	5.4 GB	38.2 mentions/s
LongBEL-8B	28.6 GB	5.4 GB	15.2 mentions/s

LongBEL has the same model memory footprint as the sentence-level Llama-8B baseline, but it is slower because it processes longer contexts and updates document-level memory during inference.

Limitations

This model assumes that mention spans and semantic groups are given. It does not perform mention detection.

LongBEL is most useful when concepts recur within a document. When most concepts appear only once, the memory mechanism has less information to exploit.

Because LongBEL uses previous predictions as memory, early mistakes can still influence later predictions. Robust memory training reduces this risk but does not remove it completely.

This model is intended for research use. It should not be used for clinical decision-making without additional validation and human oversight.

Reproducibility

Code and evaluation scripts are available in this GitHub repository.

Trained model checkpoints and processed datasets are available in the anonymous Hugging Face collection associated with LongBEL.