Symptom Analyzer: Disease Prediction System

A machine learning-based system for predicting diseases from symptoms. This project includes a training program to build machine learning models and a prediction tool for analyzing symptoms.

Features

• Model Training: Train multiple ML models (Naive Bayes, Random Forest, Deep Learning) on symptom-disease data
• Visualization: Generate visualizations of disease distributions, model performance, and feature importance
• Interactive Prediction: Use trained models to predict diseases from symptoms through command-line or interactive mode
• Confidence Levels: Get prediction confidence ratings (High/Medium/Low) to assess reliability
• Model Ensemble: Combines predictions from multiple models for more robust results

Requirements

• Python 3.6+
• scikit-learn
• numpy
• pandas
• matplotlib
• seaborn
• joblib
• TensorFlow (optional, for deep learning model)

Getting Started

Step 1: Train the models

python predict_disease.py

This will:

1. Create a sample dataset of symptoms and diseases
2. Preprocess the data and extract features
3. Train Naive Bayes and Random Forest models
4. Train a deep learning model (if TensorFlow is available)
5. Generate visualizations of model performance and feature importance
6. Save the trained models to the 'models/' directory

Step 2: Use the prediction tool

Command-line mode:

python predict_symptoms.py "fever headache cough"

Interactive mode:

python predict_symptoms.py

Follow the prompts to enter symptoms or select from example inputs.

Program Components

predict_disease.py

The main training program that:

• Creates and processes a symptom-disease dataset
• Trains multiple machine learning models
• Evaluates model performance
• Visualizes results
• Saves trained models for later use

predict_symptoms.py

The prediction tool that:

• Loads trained models
• Processes user-input symptoms
• Makes predictions using all available models
• Provides a final prediction with confidence level
• Supports both command-line and interactive modes

Example Diseases and Symptoms

The system can predict several diseases based on symptoms:

• Common Cold: fever, headache, cough, runny nose, congestion
• Heart Attack: chest pain, shortness of breath, sweating, pain radiating to arm
• Meningitis: high fever, severe headache, stiff neck, light sensitivity
• Tuberculosis: fatigue, weight loss, night sweats, persistent cough
• Gastroenteritis: abdominal pain, diarrhea, nausea, vomiting
• Diabetes: frequent urination, excessive thirst, hunger, weight loss
• Asthma: wheezing, shortness of breath, chest tightness, coughing
• Dengue: fever, rash, joint pain, pain behind eyes
• Mononucleosis: fatigue, sore throat, swollen lymph nodes

Model Performance

• Naive Bayes: Simple probabilistic classifier, typically ~85-90% accuracy
• Random Forest: Ensemble method, typically ~85-95% accuracy
• Deep Learning: Neural network approach, performance varies with dataset size

Notes

• This is a simplified demonstration system and not a substitute for professional medical advice
• Prediction accuracy depends on the quality and size of the training dataset
• Consider adding more symptom-disease examples for improved accuracy
• The confidence level helps assess the reliability of predictions

License

This project is available under the MIT License.