AID- Artificial Intelligence in Diagnosis

Inspiration

According to an article cited in the reputed newspaper, Economic Times, in November 2019, the Doctor-Patient ratio in India was 1:1445, which means that for every doctor, there were 1445 patients. Moreover, the situation is alarming because the number of specialists is a terrifyingly small number relative to the substantial number of patients. Thus, keeping this in view, we thought of developing something that can be accessible, accurate, free, and can alleviate the stress on the Healthcare system. Artificial Intelligence has been having a lot of positive influence in our lives recently because of its ability to handle huge amounts of data, come up with valuable insights, and formulate an efficient solution to pressing issues. We chose Breast Cancer and Diabetes. Breast Cancer is the most prevalent form of Cancer and Diabetes affects millions of people around the world. One of our team member's elder brother is suffering from Diabetes and it's really painful. Early diagnosis can help immensely. Moreover, in India, an average doctor sees a patient for an average of 2 minutes. So by this, we aim to provide early and accurate diagnosis which is convenient for people and they don't have to sit for hours for diagnosis.

What it does

It takes in some data from the user and generates a prediction if the user has a disease or not. The prediction is done by a Machine Learning model. The Machine Learning model for Breast Cancer has an accuracy of around 98% The Machine Learning model for Diabetes has an accuracy of around 85.7%

How we built it

The front-end was made by HTML and CSS by the help of frameworks like Bootstrap and Tailwind CSS. The Machine Learning models were integrated with the front-end by Flask and were deployed on Heroku. The domain name was provided from Domain.com.

Challenges we ran into

Collecting the datasets and building a Machine Learning model.

Accomplishments that we're proud of

We are proud of the accuracy of the Machine Learning models despite having small and medium sized datasets.

What we learned

We learned about the integrating Machine Learning model and the front-end of the website.

What's next for AID- Artificial Intelligence in Diagnosis

We aim to include more diseases like Tuberculosis, AIDS, etc. Instead of taking texts as an input, we can also use Convolutional Neural Network (CNN) for image datasets like Lung Scans. More than 90% of the medical data is in the form of images so we would have to use CNN if we want to diagnose diseases from images.