Instructions to use toolevalxm/MedDiagnosisAI-ClinicalRepo with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use toolevalxm/MedDiagnosisAI-ClinicalRepo with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="toolevalxm/MedDiagnosisAI-ClinicalRepo")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("toolevalxm/MedDiagnosisAI-ClinicalRepo")
model = AutoModelForCausalLM.from_pretrained("toolevalxm/MedDiagnosisAI-ClinicalRepo")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use toolevalxm/MedDiagnosisAI-ClinicalRepo with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "toolevalxm/MedDiagnosisAI-ClinicalRepo"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "toolevalxm/MedDiagnosisAI-ClinicalRepo",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/toolevalxm/MedDiagnosisAI-ClinicalRepo

SGLang

How to use toolevalxm/MedDiagnosisAI-ClinicalRepo with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "toolevalxm/MedDiagnosisAI-ClinicalRepo" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "toolevalxm/MedDiagnosisAI-ClinicalRepo",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "toolevalxm/MedDiagnosisAI-ClinicalRepo" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "toolevalxm/MedDiagnosisAI-ClinicalRepo",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use toolevalxm/MedDiagnosisAI-ClinicalRepo with Docker Model Runner:
```
docker model run hf.co/toolevalxm/MedDiagnosisAI-ClinicalRepo
```

MedDiagnosisAI

1. Introduction

MedDiagnosisAI represents a breakthrough in clinical decision support systems. This specialized medical language model has been fine-tuned on over 5 million de-identified clinical notes, radiology reports, and medical literature to assist healthcare professionals in diagnostic reasoning and treatment planning.

The model demonstrates exceptional performance on medical benchmarks, achieving state-of-the-art results on disease classification, clinical note summarization, and treatment recommendation tasks. On the MedQA-USMLE benchmark, MedDiagnosisAI achieves 89.2% accuracy, surpassing previous models by a significant margin.

Key improvements in this version include:

Enhanced differential diagnosis capabilities
Improved drug-drug interaction detection
Better handling of complex multi-morbidity cases

2. Evaluation Results

Comprehensive Medical Benchmark Results

	Benchmark	BioGPT	ClinicalBERT	MedPaLM	MedDiagnosisAI
Diagnostic Tasks	Disease Classification	0.721	0.745	0.812	0.899
	Symptom Extraction	0.689	0.712	0.758	0.839
	Drug Interaction	0.654	0.678	0.723	0.848
Clinical Imaging	Radiology Interpretation	0.612	0.634	0.701	0.826
	Lab Result Interpretation	0.698	0.715	0.769	0.859
	Patient History Analysis	0.723	0.741	0.789	0.860
Treatment Planning	Treatment Recommendation	0.667	0.689	0.734	0.848
	Clinical Note Summarization	0.734	0.756	0.802	0.880
	Medical QA	0.756	0.778	0.823	0.899
	Prognosis Prediction	0.645	0.667	0.712	0.846
Safety & Compliance	Adverse Event Detection	0.678	0.701	0.745	0.869
	Clinical Trial Matching	0.612	0.634	0.678	0.808
	ICD Coding	0.789	0.812	0.856	0.910
	Patient Communication	0.701	0.723	0.767	0.859
	Ethical Compliance	0.734	0.756	0.801	0.879

Overall Performance Summary

MedDiagnosisAI demonstrates superior performance across all evaluated medical benchmark categories, with particularly strong results in diagnostic and safety-critical tasks.

3. Clinical Integration & API Access

We provide HIPAA-compliant API access for healthcare institutions. Please contact our medical partnerships team for integration details.

4. How to Run Locally

Please refer to our clinical deployment guide for information about running MedDiagnosisAI in your healthcare environment.

Important usage guidelines for medical applications:

Always validate model outputs with qualified healthcare professionals.
The model is designed as a clinical decision support tool, not a replacement for medical expertise.

System Prompt for Clinical Use

We recommend the following system prompt for clinical applications:

You are MedDiagnosisAI, a clinical decision support assistant designed to help healthcare professionals.
You provide evidence-based medical information and should always recommend consultation with qualified medical professionals for final decisions.
Today is {current date}.

Temperature Settings

For medical applications, we recommend lower temperature settings:

Diagnostic queries: T=0.3
Treatment recommendations: T=0.2
General medical Q&A: T=0.5

Clinical Note Processing Template

For processing clinical notes, use this template:

clinical_note_template = \
"""[Patient ID]: {patient_id}
[Note Type]: {note_type}
[Clinical Content Begin]
{clinical_content}
[Clinical Content End]
{clinical_query}"""

5. License

This model is licensed under the Apache 2.0 License. Use in clinical settings requires additional compliance verification.

6. Contact

For clinical partnerships and research collaborations, contact us at clinical@meddiagnosisai.health

Downloads last month: 6