Kidney_failure

KO | EN

gitlite — search

#docker #framework #linux #claude #api #rag #tailwindcss #kubernetes #csharp #developer-tools

(★ 9)

No description available.

File Explorer

ZIP Download

- banner.svg
- preview.svg
- generate_data.py
- kidney_disease.csv
- kidney_model.pkl
- metrics.json
- scaler.pkl
- train_model.py
.gitignore
app.py
LICENSE
README.md
requirements.txt

# Use via CDN

jsDelivr

jsDelivr serves any public GitHub repository as a CDN with zero setup. Pick a version and a file to get a ready-to-paste link and snippet.

Version

File

Link

Example

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

A dark, gold-accented ML dashboard for estimating Chronic Kidney Disease (CKD) risk. 
 Built with Python, scikit-learn &amp; Streamlit — a deliberately balanced model wrapped in a UI that actually looks like something.

## ◆ Table of Contents

- [Features](#-features)
- [Tech Stack](#-tech-stack)
- [Requirements](#-requirements)
- [Setup](#-setup)
- [Project Structure](#-project-structure)
- [About the Model](#-about-the-model)
- [Medical Disclaimer](#-medical-disclaimer)
- [License](#-license)

## ◆ Features

| | |
|---|---|
| **✦ Balanced ML model** A `RandomForestClassifier` tuned on purpose — strong enough to be genuinely useful, capped enough to stay realistic (~85% accuracy). | **✦ 23 clinical indicators** Age, blood pressure, serum creatinine, hemoglobin, blood urea, and more — mirroring the structure of the well-known UCI CKD dataset. |
| **✦ Live risk estimation** Move the sliders, hit one button, get an instant probability with a color-coded verdict (low / moderate / high risk). | **✦ Custom dark & gold UI** A dashboard theme designed from scratch with Streamlit — no default template look. |
| **✦ Feature importance chart** See exactly which lab values are driving the model's decision for every prediction. | **✦ Reproducible pipeline** Regenerate the dataset and retrain the model with two commands — everything is scripted. |

## ◆ Tech Stack

- **Python 3.9+**
- **scikit-learn** — `RandomForestClassifier` + `StandardScaler`
- **pandas / NumPy** — data generation & processing
- **Streamlit** — the web dashboard
- **joblib** — model persistence

## ◆ Requirements

- Python 3.9 or newer
- pip
- (optional) a virtual environment tool — `venv` or `conda`

## ◆ Setup

**1 · Clone the repo**

```bash
git clone https://github.com/YOUR_USERNAME/kidney-disease-predictor.git
cd kidney-disease-predictor
```

**2 · Install dependencies**

```bash
pip install -r requirements.txt
```

**3 · (Optional) Regenerate the data & retrain the model**

The trained model is already included in `model/`, so this step is optional —
run it only if you want to reproduce or tweak the pipeline.

```bash
python data/generate_data.py
python model/train_model.py
```

**4 · Launch the dashboard**

```bash
streamlit run app.py
```

Then open **`http://localhost:8501`** in your browser.

## ◆ Project Structure

```
kidney-disease-predictor/
├── assets/
│   ├── banner.svg          → README banner
│   └── preview.svg         → README UI preview
├── app.py                  → Streamlit dashboard
├── data/
│   ├── generate_data.py    → synthetic dataset generator
│   └── kidney_disease.csv  → generated dataset
├── model/
│   ├── train_model.py      → training script
│   ├── kidney_model.pkl    → trained model
│   ├── scaler.pkl          → fitted StandardScaler
│   └── metrics.json        → accuracy & feature list
├── requirements.txt
├── LICENSE
└── README.md
```

## ◆ About the Model

The model is trained on a **synthetic dataset** whose feature structure is
inspired by the [UCI Chronic Kidney Disease dataset](https://archive.ics.uci.edu/dataset/336/chronic+kidney+disease),
though the actual values are generated, not real patient data.

The model was deliberately kept **balanced rather than maximized**:

- Limited capacity — `n_estimators=45`, `max_depth=5`, `min_samples_leaf=6`
- Statistical noise injected into the label-generating process, so the two
  classes genuinely overlap

**Result:** ~82–88% test accuracy — strong enough to be a convincing,
useful demo, without looking unrealistically perfect for a portfolio project.

## ◆ Medical Disclaimer

> ⚠️ This project is built **strictly for educational purposes**. It is trained
> on synthetic data and its predictions carry **no real diagnostic value**.
> It must never be used as a substitute for professional medical advice —
> always consult a licensed physician or nephrologist for actual health
> decisions.

## ◆ License

MIT — free to use, modify, and share.

◆ Built with Python, scikit-learn, and a little bit of gold.

Kidney_failure

Owner Repos

File Explorer

# Use via CDN

Was this content helpful?

Recent Feedback

Kidney_failure

Owner Repos

File Explorer

# Use via CDN

Was this content helpful?

Recent Feedback

Download README