Parkinson-s-Disease

🧠 Parkinson's Disease Detection using Voice Analysis

This project focuses on early detection of Parkinson’s Disease (PD) using machine learning on vocal biomarkers.
The model analyses voice features such as jitter, shimmer, and frequency variation to classify whether a patient has Parkinson’s or not — achieving 92.3% accuracy and an ROC-AUC of 0.96.

---

📊 Project Overview

Category	Details
Domain	Healthcare / Bioinformatics / AI
Objective	Early diagnosis of Parkinson’s disease using voice measurements
Algorithm Used	Random Forest Classifier
Accuracy	92.3%
ROC-AUC Score	0.962
Dataset	UCI Parkinson’s Disease Dataset
Language	Python
Libraries	Pandas, NumPy, Scikit-learn, Seaborn, Matplotlib

---

🧩 Methodology

1. Data Collection
   The dataset was obtained from the UCI Machine Learning Repository. It consists of 195 voice recordings, each with 23 biomedical voice measures.

2. Data Preprocessing

   • Dropped non-numeric columns (like patient name).
   • Scaled features using StandardScaler.

3. Model Development

   • Used RandomForestClassifier from scikit-learn.
   • Split data into 80% training and 20% testing.
   • Trained and tuned hyperparameters for optimal performance.

4. Evaluation Metrics

   • Accuracy
   • Precision, Recall, F1-Score
   • ROC-AUC Curve
   • Confusion Matrix

---

📈 Results

Metric	Score
Accuracy	0.923
Precision (PD)	0.93
Recall (PD)	0.97

🔹 Confusion Matrix

[[ 8 2] [ 1 28]]

🔹 Classification Report

Class	Precision	Recall	F1-score
0 (Healthy)	0.89	0.80	0.84
1 (Parkinson’s)	0.93	0.97	0.95

---

🧠 Insights

• Voice-based biomarkers are a non-invasive and low-cost diagnostic tool.
• Random Forest outperformed linear models in accuracy and robustness.
• The model demonstrates potential for integration into telemedicine platforms or mobile diagnostic tools.

---

🚀 How to Run

1. Clone this repository:

   git clone https://github.com/YOUR_USERNAME/Parkinson-s-Disease.git
   cd Parkinson-s-Disease
   

2. Install dependencies:

   pip install pandas numpy scikit-learn seaborn matplotlib
   

3. Run the project:

   python parkinsons_diagnosis.py
   

🧾 Future Work

• Experiment with Deep Learning models (LSTM, CNN) for audio feature extraction.
• Integrate the model with Streamlit for a web-based diagnostic tool.
• Perform feature importance analysis for interpretability.

📚 References

• Little, Max A., et al. "Exploiting Nonlinear Recurrence and Fractal Scaling Properties for Voice Disorder Detection." BioMedical Engineering OnLine, 2007.
• UCI Parkinson’s Dataset
• Scikit-learn Documentation

👩‍💻 Authors

• Aakriti Jain, Ujjawal Gaur
• B.Tech in Artificial Intelligence and Data Science
• GGSIPU, 2026

If you found this project interesting, give it a star on GitHub! 🌟