Overview / Usage

Chronic diseases like Cancer, Brain Tumors, Diabetes, etc are very dangerous and perinneal. These diseases cause deaths and is killing millions of people every year. These diseases can be cured but only if the disease is detected early. Our web application is a perfect remedy for all these problems. It uses machine learning modules set up using Random Forest Algorithm and can detect these diseases very early based on simple basic question in inputs. Our application can predict diseases like Lung Cancer, Brain Stroke and Diabetes for now with basic inputs like gender, age, blood glucode level, etc. The Machine Leaning modules used has achieved above 97% accuracy. By using our project, these chronic diseases can be detected as early as possible and an early detection of these diseases can ensure a cure for a lot of people.

Methodology / Approach

Our web application uses a front-end framework set-up using HTML and CSS. A python file runs in the back-end tht is a Machine Learning code for set up for individual diseases. The web page first asks the user about what disease they would like to predict and sends the request to the respective python file in the back-end. The inputs for the respective disease shall open according to the disease asked and inputs from the user shall be taken. Then, the Machine Learning model shall analyze based on the inputs and the output shall be a classification that says weather the user is suffering from the disease, is he prone to it or he doesn't have the disease. The Machine Learning model uses Random Forest Algorithm to do the classification and we have achieved an accuracy of above 97% for all the Machine Learning modules. This is the basic layout of our project and can be considered very useful to detect life-taking chronic diseases as earliest as possible.

Usage of Intel oneAPI

I had developed this project to detect chronic diseases before they go out of hands, that is be incurable. Using a dataset approved by WHO was not enough when the back end program couldn't give the expected performance. I had an accuracy of 92% for the brain stroke prediction with a runtime of 0.31 seconds and a similar accuracy and time for the other modules included. Using oneAPI, specifically the oneDAL library, I was able to optimize my code for better performance and efficiency.

Intel's oneAPI Data Analytics Library (oneDAL) is a library that helps speed up big data analysis by providing highly optimized algorithmic building blocks for all stages of data analytics (preprocessing, transformation, analysis, modeling, validation, and decision making) in batch, online, and distributed processing modes of computation. By using this, my time complexity of the process reduced significantly. Also, I was able to increase the accuracy to 94-96% for all the models that I used. This thus helped me to process more complex datasets that I had updated later and gave me a better performance.

Technologies Used

HTML, CSS, Python, Machine Learning, Random Forest algorithm, Data Analysis, Data Visualization, Intel oneAPI, Intel oneDAL and Intel Devcloud.

Repository

https://github.com/Jeyasundar/DC_Cetacenas

Collaborators