MIT research has used an AI model trained on cough sounds to detect an early stage COVID cough.
Identifies Cough Pattern
Those who catch COVID-19 but remain largely or completely asymptomatic (and therefore do not get a test) are proving to be a challenge in stopping the spread of the virus. With this in mind, researchers have used the knowledge that coughs can be used to help identify multiple conditions to develop an AI model that can identify the individual pattern of a COVID-19 cough and thereby, aid diagnosis.
Which Conditions?
The nature and different sounds of coughs have long been used to help give information about many different conditions. Also, recently, research published in the Lancet, which also named one of the AI COVID cough researchers as a contributor (Mar Santamaria), used automated (AI) linguistic analysis to predict future onset of Alzheimer’s.
How?
The research to identify the pattern of COVID-19 used forced-cough cell phone recordings of more than 4000 subjects to create a huge cough dataset. This dataset was then used to train an AI/machine learning model which used an adapted AI speech processing framework that could make acoustic biomarkers, thereby enabling it to tell the difference between coughs.
Results
Amazingly, when the AI model’s results were then validated with an official COVID-19 test, the model achieved COVID-19 sensitivity of 98.5 percent with a specificity of 94.2 and for asymptomatic subjects the sensitivity was 100 per cent with a specificity of 83.2 percent. This research appears to show, therefore, that it is possible for AI to pick out a COVID-19 cough from other types of cough and that, as concluded by the researchers, “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”
Next
It is understood that the research team are now working with hospitals to create an even more diverse cough dataset with plans to create an app-based diagnostic tool.
What Does This Mean For Your Business?
This shows how AI technology can be used to create unique diagnostic tools that could have a huge positive impact in tackling difficult world challenges such as the COVID-19 pandemic and how to test and screen accurately and easily. The researchers see this tool as having value in daily the screening of students, workers, and public as a way (after lockdowns) of helping businesses and transport systems to open and operate while helping to quickly (in real time) and easily and in a non-invasive way, spot spreaders and thereby potentially help economies and countries get back in control.