Multiple Disease Prediction System

A Machine Learning–Based Clinical Decision Support Framework

📌 Research Context

This repository presents a Machine Learning–based Multiple Disease Prediction System developed as part of a broader research direction in AI-driven medical diagnosis and clinical decision support systems.

The system serves as a foundational framework for investigating how machine learning models can support early disease detection, and it is designed to be extended with fuzzy logic–based reasoning and deep learning models to enhance clinical accuracy, interpretability, and uncertainty handling.

This work aligns with ongoing doctoral research in the area of hybrid intelligent systems for healthcare analytics.

---

🩺 Diseases Covered

The framework currently supports prediction for four major diseases using validated medical datasets:

• Diabetes Mellitus
• Heart Disease
• Parkinson’s Disease
• Breast Cancer

Each disease module is developed and evaluated independently, allowing comparative analysis across different machine learning techniques.

---

🧠 System Overview

The web-based application enables users to:

1. Select a disease prediction module
2. Input clinically relevant features
3. Obtain real-time prediction results

The architecture is modular and extensible, supporting future integration of:

• Fuzzy inference systems for risk stratification
• Deep learning models for enhanced predictive performance
• Hybrid ML–DL–Fuzzy decision pipelines

---

🧩 Research-Oriented Architecture

Patient Clinical Data
        ↓
Data Preprocessing & Feature Engineering
        ↓
Machine Learning Models
        ↓
Probability-Based Predictions
        ↓
(Planned Extension)
Fuzzy Logic–Based Clinical Risk Interpretation

---

📊 Datasets

Publicly available benchmark datasets were used to ensure reproducibility and comparability.

Disease	Samples	Features
Diabetes	768	8
Heart Disease	1025	14
Parkinson’s Disease	195	22
Breast Cancer	569	30

Each dataset includes clinically significant attributes such as physiological measurements, biomedical signals, and diagnostic indicators.

---

⚙️ Technologies Used

• Programming Language: Python
• Web Framework: Streamlit
• Machine Learning: Scikit-learn, XGBoost
• Data Analysis: Pandas, NumPy
• Visualization: Matplotlib, Seaborn

---

🔬 Model Development Methodology

The model development process follows a research-driven pipeline:

1. Exploratory Data Analysis (EDA)

2. Data preprocessing

   • Missing value handling
   • Outlier treatment
   • Feature scaling and encoding

3. Feature selection using Recursive Feature Elimination (RFE)

4. Training multiple classifiers per disease

5. Hyperparameter optimization using GridSearchCV

6. Model evaluation using classification metrics

This systematic approach enables transparent performance comparison and reproducibility.

---

🤖 Machine Learning Models Implemented

Diabetes Prediction

• Support Vector Classifier (SVC)
• Logistic Regression
• Random Forest Classifier

Heart Disease Prediction

• XGBoost
• Random Forest Classifier
• Logistic Regression

Parkinson’s Disease Prediction

• K-Nearest Neighbour (KNN)
• XGBoost
• Random Forest Classifier

Breast Cancer Prediction

• Logistic Regression
• XGBoost
• K-Nearest Neighbour (KNN)

---

📈 Model Evaluation Results

Diabetes Prediction

• SVC – 69.48%
• Logistic Regression – 70.13%
• Random Forest – 92.32%

Heart Disease Prediction

• XGBoost – 100%
• Random Forest – 100%
• Logistic Regression – 88.31%

Parkinson’s Disease Prediction

• KNN – 100%
• Random Forest – 94.87%
• XGBoost – 92.31%

Breast Cancer Prediction

• Logistic Regression – 97.37%
• XGBoost – 97.37%
• KNN – 96.49%

Note: These results serve as baseline benchmarks for future hybrid AI model comparisons.

---

🚀 Deployment

The system is deployed on Streamlit Cloud for demonstration and accessibility:

🔗 https://ml-multiple-diseases-predictions-shashank.streamlit.app/

---

🎯 Research Contributions

• Development of a multi-disease ML-based clinical decision support framework

• Comparative evaluation of classical ML models across multiple medical datasets

• Establishment of a baseline system for future:

  • Fuzzy logic–based diagnostic reasoning
  • Deep learning–based medical prediction
  • Hybrid intelligent healthcare systems

---

🤝 Contributing

Contributions related to:

• Intelligent healthcare systems
• Fuzzy inference models
• Deep learning extensions are welcome.

---

📬 Contact

Author: Shashank Shatraboina 📧 Email: shashankshatraboina@gmail.com

--