Disease Diagnosis Based on Symptoms

This project leverages Machine Learning (ML) and Information Retrieval (IR) techniques to detect diseases based on user-inputted symptoms. It provides detailed information about the top predicted diseases, including treatment recommendations.

Table of Contents

• Introduction
• Features
• Dataset
• Installation
• Usage
  • TF-IDF Interaction
  • Machine Learning Interaction
• Results
• Contributors
• License

Introduction

Early and accurate disease diagnosis is crucial for effective treatment. This project utilizes both ML and IR methods to predict potential diseases from a given set of symptoms, aiming to assist in early detection and provide relevant treatment information.

Features

• Symptom Input: Users can input a combination of symptoms.
• Disease Prediction: The system predicts possible diseases based on the input symptoms.
• Treatment Recommendations: Provides information on treatment options for the predicted diseases.
• Two Approaches:
  • TF-IDF Cosine Similarity: Uses term frequency-inverse document frequency and cosine similarity for disease prediction.
  • Machine Learning Models: Implements various ML algorithms for prediction.

Dataset

The dataset includes records of diseases and their associated symptoms.

Installation

1. Clone the Repository:

   git clone https://github.com/abhinavsaurabh/Disease-Diagnosis-based-on-Symptoms.git
   cd Disease-Diagnosis-based-on-Symptoms

2. Set Up a Virtual Environment (optional but recommended):

   python -m venv env
   source env/bin/activate  # On Windows: env\Scripts\activate

3. Install Dependencies:

   pip install -r requirements.txt

Usage

TF-IDF Interaction

1. Files Required:

   • combination (3).csv
   • cosine_results.py
   • generatesymptoms.py
   • Interaction_TF_IDF_Cosine.ipynb
   • normal (3).csv
   • synonyms.py
   • tf_idf_result.py
   • tokenizer.py

2. Instructions:

   • Ensure all the above files are in the same directory.
   • Open and run Interaction_TF_IDF_Cosine.ipynb sequentially in a Jupyter Notebook environment.
   • For Google Colab users, upload all the files in the same runtime session and execute the notebook.

Machine Learning Interaction

1. Files Required:

   • combinational (3).csv
   • decisionT.py
   • IRadaboost.py
   • IRgdb.py
   • IRsvm.py
   • IRxgb.py
   • knneigbh.py
   • LR.py
   • mnb.py
   • multiLP.py
   • normal (3).csv
   • randomrf.py
   • synonyms.py
   • IR_final_ML_interaction_.ipynb

2. Instructions:

   • Place all the listed files in the same directory.
   • Open and execute IR_final_ML_interaction_.ipynb sequentially in Jupyter Notebook.
   • For Colab users, upload all files in the same runtime session before running the notebook.

Results

The performance of various models is as follows:

Model	Accuracy (%)
Multi-Layer Perceptron	89.42
Decision Tree	72.62
Random Forest	89.97
Logistic Regression	89.88
K-Nearest Neighbour	89.97
Support Vector Machine	87.99
Multinomial Naive Bayes	81.30
Gradient Boosting Machine	80.94
Extreme Gradient Boosting	78.83

Contributors

• Abhinav Saurabh

License

This project is licensed under the MIT License. See the LICENSE file for details.