This model is a generative Large Language Model (based on the Qwen2.5-7B architecture) strictly fine-tuned for Vietnamese Medical Named Entity Recognition (NER). It acts as a dynamic, prompt-driven extraction agent that can accurately identify and extract any medical, clinical, or demographic entities from Vietnamese texts based on the user's instructions, outputting the results in a highly structured JSON format.

Model Details

Model Description

This model transitions away from traditional token-classification BERT models by utilizing an LLM as an intelligent extraction agent. It reads Vietnamese medical texts (such as clinical notes, user questions, or news) and outputs structured JSON lists.

The model was trained via Supervised Fine-Tuning (SFT) on a vast array of entity types across multiple datasets. By injecting the target entity schemas directly into the system prompts during SFT, the model learned the generalized skill of dynamic extraction—it extracts exactly what the prompt asks for, returning a perfectly formatted JSON schema rather than memorizing a hardcoded list of labels.

Developed by: PeterPaker123
Model type: Causal Language Model (Generative NER)
Language(s) (NLP): Vietnamese (vi)
License: Apache 2.0 (Derived from Qwen base models)
Finetuned from model: Qwen/Qwen2.5-7B-Instruct

Uses

Direct Use

Unlike traditional NER models constrained to a fixed set of predefined classes, this model supports dynamic, prompt-defined entity extraction. It acts as an intelligent agent that extracts any entity type specified in the system prompt. Beyond entity extraction, the model can evaluate the textual context of each entity to assign context flags (assertion attributes such as negation, uncertainty, severity, temporality, historical, family attribution, and conditionality).

Commonly targeted categories include (but are not limited to):

DIAGNOSIS: Confirmed diagnoses, suspected diagnoses, medical conditions, or diseases.
SYMPTOM: Subjective symptoms reported by the patient, or objective clinical signs observed by healthcare providers.
TEST_NAME: Name of a laboratory test, imaging study, physical examination, or other clinical examination.
TEST_RESULT: Numerical values, categories, or descriptions associated with an examination.
DRUG: Medication mention including brand name, generic drug name, dosage/strength, or route.
ANATOMY: Anatomical locations, organs, or body structures.
PROCEDURE: Medical procedures, surgeries, therapies, or interventions.
DEMOGRAPHICS: Patient-specific identifiers such as age, gender, occupation, or location.

The model can additionally attach context flags to each extracted entity:

is_negated: true only if the entity is explicitly denied (e.g., "không", "chưa", "phủ nhận", "loại trừ").
is_uncertain: true if the entity is suspected, possible, or hedged rather than confirmed (e.g., "nghi ngờ", "có thể", "theo dõi", "chưa loại trừ").
severity: A category describing intensity when stated (e.g., "nhẹ", "vừa", "nặng"); null if not mentioned.
temporality: Whether the entity is "current" (present/ongoing) or "recent" (a recent but resolved occurrence).
is_historical: true if the entity refers to a past occurrence (e.g., "cách đây", "tiền sử", "đã từng").
is_family: true only if the entity belongs to a family member (e.g., "bố", "mẹ", "gia đình") rather than the patient.
is_conditional: true if the entity is hypothetical or conditional rather than an actual finding (e.g., "nếu", "khi", "trong trường hợp").

How it works: You simply define the desired categories and their descriptions in the system prompt, and the model will strictly adhere to extracting only those requested types into the required JSON schema.

Downstream Use

This model can be integrated into downstream healthcare applications such as:

Medical chatbot prompt enrichment (RAG pipelines).
Automated structuring of telehealth transcripts and clinical notes.
Clinical knowledge graph construction.
Epidemiological data mining and surveillance.

Out-of-Scope Use

Medical Diagnosis or Triage: This model is strictly an extraction tool. It does not diagnose, prescribe, or provide medical advice.
General Conversation: The model has been aggressively aligned to output JSON lists. It will perform poorly at standard conversational AI tasks.

Bias, Risks, and Limitations

Clinical Disclaimer: This model is an AI research tool and must not be used for automated medical diagnosis, treatment planning, or life-or-death triage without human oversight.
Language Limitation: The model is exclusively trained for the Vietnamese language.
Hallucination Residuals: While fine-tuning significantly improves adherence to the text, Generative LLMs may occasionally alter the spelling of extracted text or hallucinate entities, especially if the input text is highly ambiguous or noisy.
Context Flag Errors: The assertion flags (is_negated, is_uncertain, severity, temporality, is_historical, is_family, is_conditional) rely on the model's interpretation of surrounding language and may be misassigned in complex, nested, or ambiguous clinical phrasing.

Recommendations

Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. Extracted entities in a clinical setting must be validated by a human medical professional or a deterministic cross-checking system.

How to Get Started with the Model

Use the code below to get started with the model.

python

import json
from transformers import AutoModelForCausalLM, AutoTokenizer

model_id = "PeterPaker123/Qwen2.5-7B-Vietnamese-Medical-NER"

tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(model_id, device_map="auto", torch_dtype="auto")

system_prompt = """You are a medical expert. Your task is to identify and extract entities from the given text.
The entity types to extract are:
- DIAGNOSIS: confirmed diagnoses, suspected diagnoses, medical conditions, or diseases.
- SYMPTOM: subjective symptoms reported by the patient (CƠ_NĂNG) or objective clinical signs observed by healthcare providers (THỰC_THỂ).
- TEST_NAME: name of a laboratory test, imaging study, physical examination, or other clinical examination.
- TEST_RESULT: numerical values, categories, or descriptions associated with an examination.
- DRUG: medication mention including brand name, generic drug name, dosage/strength, route.
- ANATOMY: anatomical locations, organs, or body structures.
- PROCEDURE: medical procedures, surgeries, therapies, or interventions.
- DEMOGRAPHICS: patient-specific identifiers such as age, gender, occupation, or location.

For each extracted entity, evaluate the text context and determine the following boolean flags:
1. "is_negated": true ONLY IF explicitly denied ("không", "chưa", "phủ nhận", "loại trừ"). Otherwise, false.
2. "is_uncertain": true if suspected, possible, or hedged rather than confirmed ("nghi ngờ", "có thể", "theo dõi", "chưa loại trừ"). Otherwise, false.
3. "severity": a category describing intensity when stated ("nhẹ", "vừa", "nặng"). null if not mentioned.
4. "temporality": "current" if present/ongoing, or "recent" if a recent but resolved occurrence.
5. "is_historical": true if it occurred in the past ("cách đây", "tiền sử", "đã từng"). Otherwise, false.
6. "is_family": true ONLY IF it belongs to a family member ("bố", "mẹ", "gia đình"). False if it belongs to the patient.
7. "is_conditional": true if hypothetical or conditional rather than an actual finding ("nếu", "khi", "trong trường hợp"). Otherwise, false.

Return the result as a JSON list containing dictionaries with "entity", "type", "is_negated", "is_uncertain", "severity", "temporality", "is_historical", "is_family", and "is_conditional" keys. If no relevant entities are found, return an empty list []."""

prompt = [
  {
    "role": "system",
    "content": system_prompt
  },
  {
    "role": "user",
    "content": "Text: Bệnh nhân nam 45 tuổi vào viện vì đau tức ngực trái dữ dội. Tiền sử tăng huyết áp, đang dùng Amlodipine 5mg. Khám phổi không ran. Theo dõi nhồi máu cơ tim. Chỉ định chụp mạch vành. Bố bệnh nhân bị đái tháo đường."
  }
]

inputs = tokenizer.apply_chat_template(prompt, return_tensors="pt", add_generation_prompt=True).to(model.device)
outputs = model.generate(inputs, max_new_tokens=768, temperature=0.0)

response = tokenizer.decode(outputs[0][inputs.shape[-1]:], skip_special_tokens=True)
print(response)
# Expected Output: 
# [
#   {"entity": "nam", "type": "DEMOGRAPHICS", "is_negated": false, "is_uncertain": false, "severity": null, "temporality": "current", "is_historical": false, "is_family": false, "is_conditional": false},
#   {"entity": "45 tuổi", "type": "DEMOGRAPHICS", "is_negated": false, "is_uncertain": false, "severity": null, "temporality": "current", "is_historical": false, "is_family": false, "is_conditional": false},
#   {"entity": "đau tức ngực trái dữ dội", "type": "SYMPTOM", "is_negated": false, "is_uncertain": false, "severity": "nặng", "temporality": "current", "is_historical": false, "is_family": false, "is_conditional": false},
#   {"entity": "ngực trái", "type": "ANATOMY", "is_negated": false, "is_uncertain": false, "severity": null, "temporality": "current", "is_historical": false, "is_family": false, "is_conditional": false},
#   {"entity": "tăng huyết áp", "type": "DIAGNOSIS", "is_negated": false, "is_uncertain": false, "severity": null, "temporality": "current", "is_historical": true, "is_family": false, "is_conditional": false},
#   {"entity": "Amlodipine 5mg", "type": "DRUG", "is_negated": false, "is_uncertain": false, "severity": null, "temporality": "current", "is_historical": false, "is_family": false, "is_conditional": false},
#   {"entity": "ran", "type": "SYMPTOM", "is_negated": true, "is_uncertain": false, "severity": null, "temporality": "current", "is_historical": false, "is_family": false, "is_conditional": false},
#   {"entity": "nhồi máu cơ tim", "type": "DIAGNOSIS", "is_negated": false, "is_uncertain": true, "severity": null, "temporality": "current", "is_historical": false, "is_family": false, "is_conditional": false},
#   {"entity": "chụp mạch vành", "type": "PROCEDURE", "is_negated": false, "is_uncertain": false, "severity": null, "temporality": "current", "is_historical": false, "is_family": false, "is_conditional": false},
#   {"entity": "đái tháo đường", "type": "DIAGNOSIS", "is_negated": false, "is_uncertain": false, "severity": null, "temporality": "current", "is_historical": false, "is_family": true, "is_conditional": false}
# ]

Training Details

Training Data

This model was trained on a consolidated corpus derived from four premier Vietnamese medical datasets.

Unlike models mapped to a fixed set of classes, this model was Supervised Fine-Tuned (SFT) on ALL available entity types across these datasets. This extensive exposure taught the model how to dynamically parse the system prompt and extract the requested target types accurately.

Dataset Name	Domain / Source	Total Samples	Entities Used During SFT
VietBioNER	Tuberculosis (TB) clinical treatment	~1,700	Symptom & Disease, Diagnostic Procedure, Location, Date
PhoNER_COVID19	COVID-19 pandemic surveillance	~10,000	Symptom & Disease, Patient ID, Age, Gender, Location, Org
ViMQ	Healthcare dialogue systems	~10,000	Symptom, Disease, Drug, Procedure, Body Structure
ViMedNER	General Medical NER	~10,000	Symptom, Disease, Drug, Body Part, Procedure, Time

Data Split: Standard Train/Validation splits were maintained to monitor formatting accuracy and exact match (F1) scores during SFT.

Training Procedure

Preprocessing

Prompts were formatted into a strict ChatML format using text-only templates.
Dynamic Prompt Injection: During SFT, training samples were dynamically generated to include varying subsets of target entities in the system prompt. This taught the model to read instructions carefully and only extract what was requested, rather than hallucinating categories from memory.

Training Hyperparameters

Training regime: bf16 mixed precision
Supervised Fine-Tuning (SFT): Full parameter fine-tuning, Max Sequence Length 4096, Batch Size 2, Gradient Accumulation 8, Learning Rate 2e-5, Cosine Scheduler. Full AdamW optimizer.

Technical Specifications

Model Architecture and Objective

Qwen2.5-7B Architecture. Causal Language Modeling objective fine-tuned to act as an entity extraction classifier capable of dynamic zero-shot categorization.

Hardware

NVIDIA RTX PRO 6000 Blackwell

Software

transformers
trl (SFTTrainer)
torch (with sdpa attention)

Citation

BibTeX:

Please cite the dataset authors who made this consolidated training possible:

bibtex

@inproceedings{vietbioner,
    title = "{A Named Entity Recognition Corpus for Vietnamese Biomedical Texts to Support Tuberculosis Treatment}",
    author = "Phan, Uyen and Nguyen, Phuong and Nguyen, Nhung",
    booktitle = "Proceedings of the 13th Language Resources and Evaluation Conference",
    year = "2022",
    publisher = "European Language Resources Association"
}

@inproceedings{PhoNER_COVID19,
    title = {{COVID-19 Named Entity Recognition for Vietnamese}},
    author = {Thinh Hung Truong and Mai Hoang Dao and Dat Quoc Nguyen},
    booktitle = {Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies},
    year = {2021}
}

@article{vimq2023,
  title={ViMQ: A Vietnamese Medical Question Dataset for Healthcare Dialogue System Development},
  author={Vu, Hieu M. and Phan, Long and Nguyen, Nhung and others},
  year={2023}
}

@article{10.4108/eetinis.v11i3.5221,
    author={Pham Van Duong and Tien-Dat Trinh and Minh-Tien Nguyen and Huy-The Vu and Minh Chuan Pham and Tran Manh Tuan and Le Hoang Son},
    title={ViMedNER: A Medical Named Entity Recognition Dataset for Vietnamese},
    journal={EAI Endorsed Transactions on Industrial Networks and Intelligent Systems},
    volume={11},
    number={4},
    publisher={EAI},
    year={2024},
    doi={10.4108/eetinis.v11i3.5221}
}

Glossary

NER: Named Entity Recognition
SFT: Supervised Fine-Tuning
Dynamic Extraction: The ability of an LLM to identify unseen or varying categorizations purely based on the instructions provided in the system prompt.
Assertion Flags: Contextual attributes (is_negated, is_uncertain, severity, temporality, is_historical, is_family, is_conditional) that qualify each extracted entity based on its clinical context.

PeterPaker123/Qwen2.5-7B-Vietnamese-Medical-NER-GRPO

Model Card for PeterPaker123/Qwen2.5-7B-Vietnamese-Medical-NER