Hey everyone!

I'm excited to share some significant progress on the PolyDisease Predictor, a @streamlit web application I've been building to predict various diseases based on user input. In this update, we've laid some crucial groundwork for future enhancements and are actively expanding the application's capabilities.

Laying the Foundation: MySQL Backend Integration

One of the key developments has been the successful integration of a MySQL backend into the PolyDisease Predictor. This marks a significant step forward, allowing the application to now persistently store user interactions and feedback across all prediction modules (Diabetes, Heart Disease, and Multiple Disease Prediction).

Why a Backend?

Implementing a database opens up a world of possibilities for improving the application, including:

  • Gathering Insights: We can now collect valuable data on how users interact with the predictor, which symptoms are frequently entered, and the accuracy of the predictions based on user feedback.
  • Model Refinement: This data can be crucial for analyzing the performance of our prediction models and identifying areas for improvement and retraining.
  • Future Features: Having a backend infrastructure in place will enable us to explore exciting new features in the future, such as personalized risk assessments and tracking disease prevalence trends.

Your Feedback Matters: Implementing a Feedback Mechanism

To make the most of this new backend, we've also implemented a simple yet powerful feedback mechanism. After receiving a prediction for diabetes, heart disease, or multiple diseases, users can now indicate whether the prediction was "👍 Correct" or "👎 Incorrect." This direct feedback is now being stored in our MySQL database, providing us with valuable signals on the application's accuracy.

Expanding Our Knowledge: Adding More Symptoms and Diseases (Work in Progress!)

The PolyDisease Predictor's core strength lies in its ability to predict multiple diseases based on user-provided symptoms. We're actively working on expanding the range of symptoms and the number of diseases the application can predict.

While this is an ongoing process, the integration of the MySQL backend provides the necessary structure to manage and utilize this growing dataset effectively. We're continuously training and refining our models with this expanded information.

Tech Deep Dive: Streamlit Secrets for Database Configuration

For those interested in contributing or running the application locally, the MySQL database connection details are managed securely using Streamlit's secrets.toml file. This allows for easy configuration without hardcoding sensitive information directly into the application.

Example secrets.toml Structure:

[mysql]
host = "your_mysql_host"
user = "your_mysql_user"
password = "your_mysql_password"

Remember to replace the placeholder values with your actual MySQL credentials and ensure this file is not committed to any public repositories (add it to your .gitignore!).

What's Next?

We're continuing to focus on:

  • Expanding the database of symptoms and diseases.
  • Analyzing the collected feedback to improve the accuracy of our prediction models.
  • Exploring new features that leverage the backend infrastructure.

Stay tuned for more updates as we continue to evolve the PolyDisease Predictor! Your feedback and contributions are always welcome.

GitHub Repository:GitHub Repository

Deployed Working Model:Deployed Model