Clicky
National, Back Page

Startup launched for automated Covid detection

‘Virufy’ collecting samples for global verification


Published : 03 Sep 2021 09:32 PM | Updated : 04 Sep 2021 12:53 AM

A group of researchers and enthusiastic medical students have undertaken a startup initiative to bring Covid testing into a smart phone application.

Virufy, based in Berkeley Institute at the University of California is an innovative healthcare initiative that developed an algorithm to be used to build a digital application that can detect Covid-19 in less than 10 seconds.

However, to make the AI (Artificial Intelligence) suitable for  real-world application, cough sounds are being collected as samples from various countries, including Bangladesh, to make the technology more sensitive and specific. 

Their founder Amil Khanzada, a data scientist who developed the algorithm claims that the presence of Covid-19 in a person's lung can be detected by analyzing the frequency of that person's cough sound. 

The idea of the app that has been developed is that a person who is feeling sick and wants to check whether he/she is infected by Covid or not can open the app and cough three times in order for the app to gather the sound samples. If the algorithm can match the patterns of the sound programmed in the app as Covid positive, it will show a result accordingly. 

The experiment has been replicated by MIT, Cambridge University and various other research facilities and all of them have come to the conclusion that this algorithm can detect the presence of the virus immediately through a digital application on the phone.

According to reports, under laboratory conditions, Virufy AI can detect Covid-19 with an accuracy of 90.4%. The process simply involves going into the app, cough 3 times and wait for the AI to analyze the cough. The result is displayed in seconds.

The global outreach heads of the startup are currently working to collect cough sound samples from around the world. When the accuracy of the AI is close to 100%, it will be released for global use.

Founder Amil Khanzada said, ‘Virufy’ is aiming to turn everyone’s smart phone into a Covid testing kit by utilizing AI to analyze a unique cough and breathing patterns of lungs affected by the virus.

He told this correspondent, “Virufy intends to work with medical systems globally to bring an end to this pandemic. What sets Virufy apart is that we are uniting the youths of the world.”

He added that it was possible due to the efforts of a young medical team including medical, technology and business experts from 15 countries and 25 universities. They are advised by Stamford Artificial Intelligence, medical professors and Silicon Valley business leaders. 

Awsaf Karim, Global Outreach Lead for Bangladesh at Virufy, is one of the healthcare entrepreneurs from 190 countries who developed this algorithm into a mobile application that anyone can install in their smart phone and use it to detect Covid-19 for free. He told Bangladesh Post that the current Covid-19 detection methods (RT-PCR and RAT) are relatively time-consuming. Moreover, it takes a long time to inform the patients that they have been tested positive. Late detection leads to late treatment which in turn results in increased death rates from Covid-19. He said, “Using our technology, anyone can detect Covid-19 in seconds from literally anywhere. Thus, they can start their treatment earlier and therefore, reduce their chances of dying from Covid-19.”

He added that they are currently collecting cough sound samples.  One can download the official app from Play Store or App store or they can visit virufy.org to donate the cough sound. Many a times, a person with Covid is not actually subjected to coughing. This correspondent asked the representative of Virufy how the app would work if coughing is not part of their symptoms.

He said, “The cough collection is just the first part of the project. In Colombia, our sister project Arkangel AI has already developed the ability to interpret CT scan and X-ray reports. So, a fast result could be determined by our app from the reports as well”.

He added that they have launched a GoFundMe and hope that once the cough-based testing is successful, they will receive funding to develop the system to detect asymptomatic cases.

Virologist Dr Saif Ullah Munshi told this correspondent that he is unsure how effective this may be. However, technological advancements could be surprising.

“I do not know if this is possible. However, a lot of research is in works and this could be a development yet to come,” he said. Tech experts believe that if a pattern can be matched, an algorithm as such can be developed and AI can effectively give out results. However, the pattern needs to be successfully verified for all regions.  When asked about plans to bring Virufy to Bangladesh, officials said that first and foremost, the idea is to make the app suitable for this region (e.g., by translating it into Bangla). Secondly, cough sound samples need to be collected to help increase the accuracy of the AI. Their aim is to pitch the launch though the backing of renowned doctors and pharmaceutical companies so that they could confirm the efficiency of the initiative. Finally, findings and proposal need to be submitted to the government to apply for official certifications permission from the health sector of the country.