Authors

Inspiration

To prevent the COVID-19 infections spreading, lots of diagnosis test-kits have been developed all around the world. Diagnosis is important for the isolation of the infected people, preventing them from spreading the disease and is also important for the observation and study of the infection. Most of the test kits based on the samples taken from blood, mucosa layer inside the nose and throat. These samples are used in RT-qPCR, antigen and antibody tests. RT-qPCR technique is the one with highest confidence, however it is costly, not portable and takes too much time. The problems with RT-qPCR raise the need for cheaper, efficient and portable new methods.

According to WHO, *dry cough is an important symptom for 67,7% of the patients. *WHO A dry cough is when the cough does not include mucus secretion. Coughing of people with COVID-19 can have a distinct sound profile that can be differentiated from other coughing sounds. The different sound structure can be analyzed by algorithms we have developed, and these algorithms can be used as a novel diagnosis tool for COVID-19 that is cheaper, efficient and portable.

The Aim

This research aims to build up machine learning algorithms that can inform the diagnosis of COVID-19 based on the people's coughing and breathing sounds. After developing an algorithm by using large scale, crowdsourced data collection from healthy and non-healthy participants, we can create a novel diagnostic tool for COVID-19 that is cheaper, efficient and portable and we can use it as an early diagnostic tool.

If successful, all the countries and people in the world will be the beneficiaries. Moreover, the work in this project can also be scaled for the other respiratory diseases rather than COVID-19.

ML Model

This is the research repository of Covid-19 Cough Sound Classification. You can read the details of research from the web site we prepared.

https://caglayan.github.io/covid19/en/research

Team

Our team is made up of two graduate students from Boğaziçi University, Istanbul.

Çağlayan Şerbetçi is a graduate student in System and Control Engineering and M. Görkem Durmaz has a bachelor’s degree in Molecular Biology & Genetics and she is now a graduate student in Computer Engineering department.

Project Web Site

In this project, COUGH AI, we have developed a website. This web site record coughs with various diagnostic information. It works on all mobile devices. Please record your cough sounds and help our research!

Web site: https://caglayan.github.io/covid19/en/ more details.

Try It out

Hackathons

Technologies

javascript, machine-learning, matlab, node.js, react

Devpost Software Identifier

267123