• 

  DiabTech Logo

DiabTech: Revolutionizing Diabetes Prediction

Inspiration

DiabTech was born out of a profound inspiration stemming from the challenges faced by our own family members with diabetes. Motivated by the opportunity to make a meaningful impact, our team embarked on a journey to harness the power of the latest machine learning (ML) and deep learning models to create an accessible and user-friendly diabetes prediction tool.

What it does

DiabTech is at the forefront of predictive modeling for Type-2 Diabetes, featuring a cutting-edge ensemble of a Random Forest Model and a Multi-Layer Perceptron Neural Network. These models are seamlessly integrated into a robust Full Stack Web Application, leveraging the strengths of React, Django, and SQLite. Our primary objective is to offer a smooth, user-friendly interface for effective diabetes risk prediction and management.

How we built it

The project was divided into two key components:

1. Machine Learning/Deep Learning: Utilizing powerful tools such as Python, PyTorch, Google Collaboration, and scikit-learn, we trained our classification model and fine-tuned its performance.
2. Web Development: Setting up the backend using Python and the Django framework, we incorporated technologies like React and Material UI to ensure a responsive and feature-rich frontend.

Challenges we ran into

• Learning Django, Python, React, Redis Cloud, Python and Material UI from scratch.
• Implementing Redis Cloud as a cache using the Redis stack to enhance the web application's speed.
• Balancing a dataset of approx. 100,000 data points.
• Addressing challenges related to overfitting and model tuning.

Accomplishments that we're proud of

1. Development of a Practice Research Paper.
2. Implementation of a Cache using Redis Cloud technology, reducing computation time for redundant data by up to ten times.
3. Successful training of a Random Forest Ensemble model and Multi-Layer Perceptron Neural networks using a dataset of approximately 100,000 patients with and without Type 2 Diabetes.

What we learned

• The importance of efficient task delegation based on team members' strengths.
• The significance of leveraging technology to make healthcare more accessible and inclusive.

What's next for DiabTech

• Introducing a live training feature for user data to continually enhance the model's learning capabilities.
• Expanding predictive software to address other medical conditions.
• Exploring scalability through containerization and cloud deployment.

Built With

• bootstrap
• cache
• cloud
• css3
• django
• googlecollab
• html5
• javascript
• materials
• python
• pytorch
• react
• redis
• skicit-learn
• sqlite