OncoPredict AI: Breast Cancer Diagnostic System 🩺

A machine learning-powered clinical decision support system that predicts breast cancer malignancy based on cell nuclei measurements. Built with Python, Scikit-Learn, and Streamlit.

(Add a screenshot of your beautiful UI here after you run it!)

💡 About The Project

Early diagnosis is the most critical factor in breast cancer survival rates. OncoPredict AI is a bridge between raw medical data and actionable clinical insights.

Instead of leaving the machine learning model inside a Jupyter Notebook, I deployed it as an interactive web application. This tool allows medical professionals (or users) to input specific cytology features—such as radius, texture, and smoothness—and receive an instant, probability-based prediction on whether a tumor is Benign or Malignant.

The goal was to build a system that is not only accurate but also visually accessible, moving away from complex code interfaces to a clean, professional dashboard.

⚙️ How It Works (The ML Pipeline)

This project follows a rigorous Data Science lifecycle:

1. Data Ingestion: Utilizes the Wisconsin Breast Cancer Diagnostic (WBCD) dataset.
2. Preprocessing: * Cleaned the data by removing irrelevant IDs and empty columns.
   • Mapped categorical diagnosis (M/B) to binary targets (1/0).
3. Scaling: Implemented a StandardScaler to normalize feature values (e.g., ensuring 'Area' doesn't dominate 'Smoothness' just because the numbers are larger).
4. Modeling: Trained a Logistic Regression classifier, optimized for binary classification tasks in medical contexts.
5. Deployment: Serialized the model and scaler using Pickle and built a frontend with Streamlit.

✨ Key Features

• Dynamic Input System: The app automatically reads the dataset features and generates the appropriate sliders, making the code adaptable to different datasets.
• Real-Time Probability Score: Doesn't just say "Cancer" or "Safe"; it provides a confidence percentage (e.g., "98.5% Confidence").
• Professional UI: Custom CSS implementation for a "Glassmorphism" look, featuring clean typography and specific color coding (Green for Benign, Red for Malignant).
• Stateful Persistence: Uses pickle to load trained assets instantly without retraining on every reload.

🛠️ Tech Stack

• Language: Python 3.10+
• Frontend: Streamlit (with Custom CSS)
• ML Libraries: Scikit-Learn, Pandas, NumPy
• Serialization: Pickle

🚀 How to Run Locally

1. Clone the Repository

git clone [https://github.com/yourusername/oncopredict-ai.git](https://github.com/yourusername/oncopredict-ai.git)
cd oncopredict-ai