After close to a year of living with the coronavirus, it seems that many people are no longer stressing out after every sneeze or cough, or when their throat tickles. But there are still many out there who worry about catching the virus, and with the U.S. ringing up 90,000-100,000 new cases a day and Britain on lockdown for the second time, this is certainly not the time to declare victory and stop wearing masks.
And if you still reach for the thermometer every time you cough, sneeze, or swear that you can’t taste that spicy burrito you ordered from Chipotle, soon there might be an app that will give you a definitive answer to the question, “Do I have COVID?” A report issued by the Institute of Electrical and Electronics Engineers (IEEE)
discusses its use of Machine Learning to analyze coughs. Using ML, the app learns to tell the difference between those who have COVID-19 but are asymptomatic, and those who are healthy and do not have the virus.
The team working on the app was able to collect 70,000 recordings from a website that allowed the public to leave samples of coughs over smartphones and other devices. Overall there were 200,000 forced-cough samples collected with 2,500 confirmed to have COVID or were asymptomatic. 98.5% of the time, the model was able to accurately determine whether someone had COVID while for asymptomatic subjects it achieved an accuracy rate of 100%. The conclusion reached in the report says, “AI techniques can produce a free, non-invasive, real-time, any-time, instantly distributable, large-scale COVID-19 asymptomatic screening tool to augment current approaches in containing the spread of COVID-19. Practical use cases could be for daily screening of students, workers, and public as schools, jobs, and transport reopen, or for pool testing to quickly alert of outbreaks in groups.”
The researchers are reportedly considering the distribution of a free pre-screening app that would be based on AI while at the same time working with hospitals to expand the pool of cough recordings for future testing. The models based on coughs could end up in smart speakers and digital assistants like Siri, Google Assistant and Alexa for daily analysis. For the accuracy needed to make the technology worthwhile to use, the testing would have to be done on devices with high quality microphones. Privacy issues will need to be ironed out.
Being able to open an app, ask a subject to cough, and quickly get a response about whether he or she has COVID would be a big help in trying to eradicate the disease.