{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Sybil - Lung Cancer Risk Prediction\\n",
    "\\n",
    "This notebook demonstrates how to use the Sybil model from Hugging Face for lung cancer risk prediction."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 1. Install Requirements"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "!pip install huggingface-hub torch torchvision pydicom sybil requests"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 2. Load Model from Hugging Face"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from huggingface_hub import snapshot_download\\n",
    "import sys\\n",
    "\\n",
    "# Download model\\n",
    "print(\"Downloading Sybil model from Hugging Face...\")\\n",
    "model_path = snapshot_download(repo_id=\"Lab-Rasool/sybil\")\\n",
    "sys.path.append(model_path)\\n",
    "\\n",
    "# Import model\\n",
    "from modeling_sybil_wrapper import SybilHFWrapper\\n",
    "from configuration_sybil import SybilConfig\\n",
    "\\n",
    "# Initialize\\n",
    "config = SybilConfig()\\n",
    "model = SybilHFWrapper(config)\\n",
    "print(\"✅ Model loaded successfully!\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 3. Download Demo Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import requests\\n",
    "import zipfile\\n",
    "from io import BytesIO\\n",
    "import os\\n",
    "\\n",
    "def get_demo_data():\\n",
    "    cache_dir = os.path.expanduser(\"~/.sybil_demo\")\\n",
    "    demo_dir = os.path.join(cache_dir, \"sybil_demo_data\")\\n",
    "    \\n",
    "    if not os.path.exists(demo_dir):\\n",
    "        print(\"Downloading demo DICOM files...\")\\n",
    "        url = \"https://www.dropbox.com/scl/fi/covbvo6f547kak4em3cjd/sybil_example.zip?rlkey=7a13nhlc9uwga9x7pmtk1cf1c&dl=1\"\\n",
    "        response = requests.get(url)\\n",
    "        \\n",
    "        os.makedirs(cache_dir, exist_ok=True)\\n",
    "        with zipfile.ZipFile(BytesIO(response.content)) as zf:\\n",
    "            zf.extractall(cache_dir)\\n",
    "    \\n",
    "    # Find DICOM files\\n",
    "    dicom_files = []\\n",
    "    for root, dirs, files in os.walk(cache_dir):\\n",
    "        for file in files:\\n",
    "            if file.endswith('.dcm'):\\n",
    "                dicom_files.append(os.path.join(root, file))\\n",
    "    \\n",
    "    print(f\"Found {len(dicom_files)} DICOM files\")\\n",
    "    return sorted(dicom_files)\\n",
    "\\n",
    "# Get demo data\\n",
    "dicom_files = get_demo_data()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 4. Run Prediction"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Run prediction\\n",
    "print(\"Running lung cancer risk prediction...\")\\n",
    "output = model(dicom_paths=dicom_files)\\n",
    "risk_scores = output.risk_scores.numpy()\\n",
    "\\n",
    "# Display results\\n",
    "print(\"\\n\" + \"=\"*40)\\n",
    "print(\"Lung Cancer Risk Predictions\")\\n",
    "print(\"=\"*40)\\n",
    "\\n",
    "for i, score in enumerate(risk_scores):\\n",
    "    risk_pct = score * 100\\n",
    "    bar_length = int(risk_pct * 2)  # Scale for visualization\\n",
    "    bar = '█' * bar_length + '░' * (30 - bar_length)\\n",
    "    print(f\"Year {i+1}: {bar} {risk_pct:.1f}%\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 5. Visualize Risk Progression"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt\\n",
    "import numpy as np\\n",
    "\\n",
    "# Create visualization\\n",
    "years = np.arange(1, 7)\\n",
    "risk_percentages = risk_scores * 100\\n",
    "\\n",
    "plt.figure(figsize=(10, 6))\\n",
    "plt.bar(years, risk_percentages, color=['green', 'green', 'yellow', 'yellow', 'orange', 'orange'])\\n",
    "plt.xlabel('Years from Scan', fontsize=12)\\n",
    "plt.ylabel('Lung Cancer Risk (%)', fontsize=12)\\n",
    "plt.title('Predicted Lung Cancer Risk Over Time', fontsize=14, fontweight='bold')\\n",
    "plt.grid(axis='y', alpha=0.3)\\n",
    "\\n",
    "# Add value labels on bars\\n",
    "for i, (year, risk) in enumerate(zip(years, risk_percentages)):\\n",
    "    plt.text(year, risk + 0.5, f'{risk:.1f}%', ha='center', fontweight='bold')\\n",
    "\\n",
    "plt.tight_layout()\\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 6. Using Your Own Data\\n",
    "\\n",
    "To use your own CT scan data, replace the demo data with your DICOM file paths:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Example with your own data (uncomment and modify)\\n",
    "# my_dicom_files = [\\n",
    "#     \"/path/to/your/scan/slice001.dcm\",\\n",
    "#     \"/path/to/your/scan/slice002.dcm\",\\n",
    "#     # ... add all slices\\n",
    "# ]\\n",
    "# \\n",
    "# output = model(dicom_paths=my_dicom_files)\\n",
    "# my_risk_scores = output.risk_scores.numpy()\\n",
    "# \\n",
    "# for i, score in enumerate(my_risk_scores):\\n",
    "#     print(f\"Year {i+1}: {score*100:.1f}% risk\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Important Notes\\n",
    "\\n",
    "⚠️ **Medical Disclaimer**: This model is for research and educational purposes. Always consult qualified healthcare professionals for medical decisions.\\n",
    "\\n",
    "📚 **Citation**: If you use this model in research, please cite:\\n",
    "```\\n",
    "Mikhael, P.G., Wohlwend, J., Yala, A. et al. (2023).\\n",
    "Sybil: A validated deep learning model to predict future lung cancer risk\\n",
    "from a single low-dose chest computed tomography.\\n",
    "Journal of Clinical Oncology, 41(12), 2191-2200.\\n",
    "```"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.0"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}