ML-Diagnosis

A Python GUI App Predicting Medical Diagnosis Based on User Symptoms Using Naive Bayes Classifier

Overview

ML-Diagnosis is a Python-based graphical user interface (GUI) application designed to predict medical diagnoses based on user-provided symptoms. This application leverages a Naive Bayes classifier trained on real medical data to provide the most probable diagnosis.

Motivation

My dual interest in biology and programming culminated in the development of the Mini Medical Center project, which includes ML-Diagnosis and a pill dispenser device with my team.

Project Components

1. AI Diagnosis Application

The AI diagnosis application is the core component of the Mini Medical Center project. It includes the following features:

• Symptom Input: Users can input their symptoms through a user-friendly GUI.
• Prediction Algorithm: The application utilizes a Naive Bayes classifier to predict the most likely medical diagnosis based on the input symptoms.
• Training Data: The classifier is trained on a dataset collected from hospitals, ensuring the predictions are based on real-world medical data.

2. Pill Dispenser Device

The pill dispenser is a hardware component designed to dispense pills based on a timer set by the user. Key features include:

• Timer Functionality: Users can set a timer for pill dispensing, ensuring timely medication management.
• Integration with AI Application: The dispenser can work in conjunction with the AI diagnosis application to suggest medication based on the diagnosis.

Technical Details

AI Model

• Algorithm: Naive Bayes classifier from the scikit-learn library
• Training Data: Real medical data from hospitals
• Data Processing: Utilized the pandas library for data processing and preparation
• Features: Symptoms input by users
• Labels: Medical diagnoses
• Implementation Steps:
  1. Data Cleaning: Cleaned and preprocessed the dataset using pandas.
  2. Feature Extraction: Extracted relevant features from the symptoms data.
  3. Model Training: Trained the Naive Bayes classifier using the processed data.
  4. Evaluation: Evaluated the model's performance using standard metrics (e.g., accuracy, precision, recall).

GUI

• Framework: PyQt
• Design: An event-based basic window with minimal components, implemented in Python. The GUI allows users to enter symptoms and receive a diagnosis prediction and also allows the user to set 2 timers for each of the pill present in the device's storage.

Pill Dispenser

• Hardware: Arduino UNO R3
• Programming: Arduino programming language
• Communication: The Arduino communicates with the GUI application using the PySignal framework and a USB connection.
• Functionality: The dispenser releases pills based on a timer set by the user through the GUI.