Diagnosis at Home

Project thumbnail
Correlation matrix
The diabetes risk assesor

Inspiration

November 14 marks World Diabetes Day. It is a group of diseases that result in too much sugar in the blood. The Center of Disease Control and Prevention states that there are two types: Type 1 and Type 2 diabetes. Type 2 Diabetes chronic conditions can be prevented by staying active and eating healthy food.

What it does

This is Automatic Confirmatory System identifies if a person has diabetes based on medical input provided by using Machine Learning. Based on the results the person gets tips on becoming fitter.

Note: This project is not meant to be replacement for medical diagnosis from authorised professionals.

How we built it

We build backend and frontend in the project. Backend: Exploring, Cleaning and Transforming dataset from UCI Repository. Trained Machine Learning classification models using Decision Tree Classifier, Random Forest, Gradient Boosting classifier Support Vector and K Nearest Number algorithms. We tuned each model to reduce overfitting. Then we test the models and found out that the Random Forest classifier model gave maximum accuracy. This trained model is saved and made an API endpoint to connect with front end of the website.

Frontend:

Challenges we ran into

We started the project late in the hackathon and had difficulty integrating the front end with the back end. We overcame these challenges by helping each other and learning new skills.