RoleBox Medical Advisor (Dr. Pill Goodfeeling)

A specialized medical information advisor AI trained to provide general health information, explain medical concepts, and answer common health-related questions.

⚠️ IMPORTANT MEDICAL DISCLAIMER

THIS MODEL IS NOT A SUBSTITUTE FOR PROFESSIONAL MEDICAL ADVICE

NOT for diagnosis: This model cannot diagnose medical conditions
NOT for treatment: Do not use for medical treatment decisions
NOT for emergencies: Call emergency services (911) for urgent medical situations
Consult professionals: Always consult qualified healthcare providers for medical advice
General information only: Responses are educational and informational only

Model Details

Model Description

This is a LoRA adapter fine-tuned on top of Qwen 2.5 Coder 1.5B Instruct to create a specialized medical information advisor. The model provides general health information, explains medical terminology, and answers common medical questions based on publicly available medical knowledge.

Developed by: RoleBox Team
Model type: Causal Language Model (LoRA adapter)
Language(s): English
License: Apache 2.0
Finetuned from model: Qwen/Qwen2.5-Coder-1.5B-Instruct

Model Sources

Parent Model: https://huggingface.co/Qwen/Qwen2.5-Coder-1.5B-Instruct

Uses

Direct Use

This model is designed to provide general medical information. It can:

Explain medical terminology and concepts
Provide general information about common conditions
Answer questions about symptoms (general information only)
Explain basic treatment approaches (educational purposes)
Discuss preventive health measures
Explain how medications generally work (not prescriptions)

Example Usage

from transformers import AutoModelForCausalLM, AutoTokenizer
from peft import PeftModel
import torch

# Load base model
base_model = AutoModelForCausalLM.from_pretrained(
    "Qwen/Qwen2.5-Coder-1.5B-Instruct",
    torch_dtype=torch.float16,
    device_map="auto"
)

tokenizer = AutoTokenizer.from_pretrained("Qwen/Qwen2.5-Coder-1.5B-Instruct")

# Load LoRA adapter
model = PeftModel.from_pretrained(base_model, "hmtr/rolebox.dr-pill-goodfeeling")

# Generate response
prompt = """### Instruction:
You are a medical advisor. Answer the user's question.

### User Question:
What is hypertension and how is it managed?

### Response:
"""

inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
outputs = model.generate(**inputs, max_new_tokens=256, temperature=0.7)
response = tokenizer.decode(outputs[0], skip_special_tokens=True)
print(response)

Downstream Use

This adapter can be integrated into:

Health information applications
Medical education platforms
Patient education tools
General health Q&A systems
Healthcare chatbots (for general information only)

Out-of-Scope Use

This model MUST NOT be used for:

Medical diagnosis - Only licensed healthcare providers can diagnose
Treatment recommendations - Cannot replace professional medical advice
Prescription advice - Never use for medication decisions
Emergency medical situations - Call emergency services immediately
Mental health crisis intervention - Contact crisis hotlines or emergency services
Replacing doctor visits - Always consult healthcare providers
Legal/medical liability decisions - Requires professional judgment
Personalized medical advice - Cannot account for individual health history

Critical Safety Warnings

When to Seek Professional Help

Seek immediate medical attention for:

Chest pain or pressure
Difficulty breathing
Severe bleeding
Loss of consciousness
Severe allergic reactions
Stroke symptoms (FAST: Face drooping, Arm weakness, Speech difficulty, Time to call 911)
Any medical emergency

Always consult healthcare providers for:

New or worsening symptoms
Medication questions or changes
Chronic condition management
Pregnancy-related concerns
Mental health issues
Any health concern requiring professional evaluation

Bias, Risks, and Limitations

Medical Limitations

No clinical expertise: Model has no real medical training or clinical experience
Cannot examine patients: Lacks ability to perform physical examinations or tests
No access to medical records: Cannot review individual patient history
General information only: Cannot provide personalized medical advice
May be outdated: Medical knowledge evolves; information may not reflect latest research
No liability: Not responsible for medical outcomes from using this model

Technical Limitations

Training data bias: Based on publicly available medical Q&A data
May not cover all conditions: Limited to topics in training data
English only: Currently trained only on English-language medical content
Context limitations: Cannot maintain complex multi-turn medical consultations
No verification system: Responses are not verified by medical professionals

Potential Risks

Misinterpretation: Users may misunderstand general information as personal advice
Delayed care: Users might delay seeking professional help
Incorrect information: Model may occasionally provide inaccurate information
Over-reliance: Users might rely too heavily on AI instead of professionals
False reassurance: General information might incorrectly reassure about serious conditions

Recommendations

For Users:

✅ Use for general health education only
✅ Verify all information with healthcare providers
✅ Seek professional help for any health concerns
✅ Call emergency services for urgent situations
✅ Understand this is NOT medical advice
❌ Do NOT use for diagnosis or treatment
❌ Do NOT delay professional care based on responses
❌ Do NOT make medical decisions without consulting doctors

For Developers:

Display clear medical disclaimers prominently
Implement emergency contact information (911, crisis hotlines)
Add warnings for serious symptoms
Include "consult your doctor" reminders
Monitor for misuse or harmful applications
Consider human oversight for medical content

Training Details

Training Data

The model was fine-tuned on a curated dataset of 40,644 medical question-answer pairs covering:

Common medical conditions
Symptoms and their meanings
General treatment approaches
Preventive health measures
Medical terminology
Medication information (general)
Health and wellness topics
Basic anatomy and physiology

Data sources: Publicly available medical Q&A datasets (not patient data)

Training Procedure

Fine-tuning method: LoRA (Low-Rank Adaptation)

Training Hyperparameters

Base model: Qwen/Qwen2.5-Coder-1.5B-Instruct
Training regime: fp16 mixed precision
LoRA rank (r): 16
LoRA alpha: 32
LoRA dropout: 0.05
Target modules: q_proj, k_proj, v_proj, o_proj
Number of epochs: 3
Batch size: 4
Gradient accumulation steps: 2 (effective batch size: 8)
Learning rate: 2e-4
Max sequence length: 384 tokens
Optimizer: AdamW
Training examples: 40,644

Speeds, Sizes, Times

Adapter size: ~17.5 MB
Training time: ~2-3 hours on Google Colab T4 GPU
Training platform: Google Colab (free tier)
GPU: NVIDIA Tesla T4 (16GB VRAM)
Trainable parameters: ~4.4M (0.28% of base model)

Evaluation

This model has not undergone formal medical validation or clinical trials. Responses should be verified by healthcare professionals.

Testing Data

General medical Q&A examples covering diverse topics:

Common conditions and symptoms
Treatment information
Preventive care
Health education

Metrics

Qualitative assessment of response accuracy
No clinical validation performed
No peer review by medical professionals

Regulatory & Ethical Considerations

Not a Medical Device

This model is NOT regulated as a medical device
NOT cleared by FDA or other regulatory bodies
NOT intended for clinical use
NOT validated for patient care

Privacy

Model does not store or transmit user conversations
No patient data was used in training
Users should not share sensitive health information

Liability

RoleBox Team assumes no liability for medical outcomes
Users assume all risks of using this model
Always consult licensed healthcare providers

Environmental Impact

Training was performed on Google Colab's free tier GPU infrastructure.

Hardware Type: NVIDIA Tesla T4 GPU
Hours used: ~2-3 hours
Cloud Provider: Google Cloud Platform
Compute Region: US (variable)
Carbon Emitted: ~0.15-0.20 kg CO2eq (estimated)

Technical Specifications

Model Architecture and Objective

Architecture: Transformer-based causal language model with LoRA adapters
Objective: Causal language modeling (next token prediction)
Adapter method: LoRA (Low-Rank Adaptation)
Parameter efficiency: Only 0.28% of parameters are trainable

Compute Infrastructure

Hardware

Training: Google Colab T4 GPU (16GB VRAM)
Inference: Can run on consumer GPUs (4GB+ VRAM) or CPU

Software

Framework: PyTorch
Libraries:
- Transformers (Hugging Face)
- PEFT (Parameter-Efficient Fine-Tuning)
- Accelerate
- Datasets

Citation

BibTeX:

@misc{rolebox-medical-advisor,
  title={RoleBox Medical Advisor: LoRA-finetuned Qwen 2.5 Coder for Medical Information},
  author={RoleBox Team},
  year={2025},
  publisher={HuggingFace},
  url={https://huggingface.co/hmtr/rolebox.dr-pill-goodfeeling},
  note={NOT FOR MEDICAL DIAGNOSIS OR TREATMENT}
}

Emergency Contacts

In case of medical emergency:

US Emergency: 911
Poison Control: 1-800-222-1222
Suicide Prevention: 988
Crisis Text Line: Text "HELLO" to 741741

Model Card Authors

RoleBox Team

Model Card Contact

Email: hi@rolebox.app
Website: https://rolebox.app

Framework Versions

PEFT 0.17.1
Transformers 4.48+
PyTorch 2.6+
Python 3.10+

REMINDER: This is an AI model for general information only. Always consult qualified healthcare professionals for medical advice, diagnosis, and treatment.

Downloads last month: -

Model tree for hmtr/rolebox.dr-pill-goodfeeling

Base model

Qwen/Qwen2.5-1.5B

Finetuned

Qwen/Qwen2.5-Coder-1.5B

Finetuned

Qwen/Qwen2.5-Coder-1.5B-Instruct

Adapter

(39)

this model