COVID-19 has affected over 210 countries so far and health workers are helpless at the moment in some places due to limited resources available to them. Health ministries are globally facing a shortage of medical equipment required such as testing kits, quarantine space, ventilators, etc. Many hospitals are offloading screening of potential victims of the virus and sending them to third parties.
Screening websites like the one from Alphabet’s Verily, have risen to help in the matter. One research team is using AI to test for a simple symptom usually checked by medical professionals: listening to coughing sounds of patients and then determining whether they are due for a coronavirus test or not.
This method sounds odd at first but it has data supporting it. World Health Organization has said that dry coughing is one of COVID-19’s prime symptoms and around 67% of those with dry cough were found to be infected. However, the other 32% and asymptomatic people can also be taken into account but it is better than having no clue.
It is a project of Professor David Atienza from École Polytechnique fédérale de Lausanne’s Embedded Systems Laboratory. His team is using artificial intelligence for distinguishing between different kinds of coughs, the same way doctors do. Their main goal is to reduce the number of people going to doctors demanding a test without exhibiting the right symptoms.
Coughvid can run on any web browser or device with a mic. The basic idea for the app seems to be simple. Patients cough at their mic safely and AI will determine whether their cough indicates they have COVID-19 or not.
The website has reported over 70% accuracy in all its tests. However, the technology is still under process and is learning more and more with the increasing number of people trying it out. The website also provides patients with a guide to how they can safely test their cough.