By Chathushka Perera
New York, USA (CWBN)_ The biggest challenge in the containment and management of the Covid-19 Pandemic is the lack of effective and versatile testing methods. Currently, the best means of diagnosing the infection is by conducting swab testing, also known as PCR testing, which has an accuracy between 60 to 80%.
However, a team from the Massachusetts Institute of Technology (MIT) have come up with an AI solution with a smart phone link to it, where you are only required to force a cough to find out whether you are infected or not. The AI uses the sound of a cough and attempts to recognise minor changes in them, which cannot be heard by a human ear cannot distinguish.
The development is based on Alzheimer’s detection research, involving a neural network referred to as ResNet50, which is trained with over 1000 hours of human speech, data from words spoken in various emotional states and a cough data base.
The AI automatically filters out the healthier coughs using the training modules, and has reportedly been able to identify an infection in 97.1% of 2500 Covid patients and 100% of those who were asymptomatic. However the system is not meant to be used on people already showing Covid symptoms.
The researches will be continuing to develop and test the design using more diverse data resources and identify other factors associated with detection rates. The positivity behind the likelihood of producing a mobile app has the research team thinking about privacy implications and policy already.
