🧠 Thyroid Cancer Risk & Diagnosis Prediction (Streamlit App)

This repository contains a full machine learning pipeline and an interactive Streamlit application to predict:

• 🩺 Diagnosis of Thyroid Cancer – Benign or Malignant (Binary Classification)
• 🎯 Risk Level – Low, Medium, or High (Multiclass Classification)

The project is built using clinical and medical data, and designed for both educational and practical demonstration purposes.

🧠 Models Overview

Task	Model File	Type	Classes
Diagnosis Prediction	diagnosis_predictor.pkl	Binary Classifier	Benign, Malignant
Risk Prediction	risk_predictor.pkl	Multiclass Classifier	Low, Medium, High

📁 Model & Preprocessing Artifacts (/models/)

These .pkl files are required to run the Streamlit app and are included in this repository:

🧪 Diagnosis Prediction Files:

File	Description
diagnosis_predictor.pkl	Trained binary classification model to predict benign/malignant diagnosis
scaler_diagnosis.pkl	StandardScaler used to scale numeric input features for diagnosis model
feature_columns_diagnosis.pkl	List of feature names used for diagnosis prediction inputs in the app
binning_edges_diagnosis.pkl	Binning thresholds for numerical features used in Streamlit select boxes

📊 Risk Prediction Files:

File	Description
risk_predictor.pkl	Trained multiclass model to predict thyroid cancer risk (Low/Medium/High)
scaler_risk.pkl	StandardScaler used for numeric inputs of the risk model
feature_columns_risk.pkl	List of features used in risk prediction input
binning_edges_risk.pkl	Binning thresholds used for categorical-like dropdowns in the app

📊 Dataset

The models were trained on a clinical dataset containing patient health parameters and diagnostic labels.
If you'd like to train the models yourself, you can download the dataset from:

🔗 []Kaggle Dataset Link :] (https://www.kaggle.com/datasets/bhargavchirumamilla/thyroid-cancer-risk-dataset)

🛠 Technologies Used

• Python 3.10+
• Streamlit – for building interactive web apps
• Scikit-learn – machine learning models and preprocessing
• Pandas, NumPy – data analysis and manipulation
• Joblib – efficient model serialization
• Google Colab – preprocessing and model training

📜 License

This project is open-source and free to use for non-commercial and academic purposes.
If you use this work in research or development, please provide attribution.

🙌 Acknowledgments

• Kaggle: Thyroid Disease Dataset – for providing the publicly available clinical data
• The open-source contributors to Streamlit, Scikit-learn, and XGBoost
• The broader Python and machine learning communities for open knowledge sharing