A local engineering services company has developed an Artificial Intelligence (AI) algorithm that identifies people with COVID-19 just by the sound of cough.

The Serba Dinamik Artificial Intelligence Team (SDAI), led by Artificial Intelligence Society Malaysia (ARTIS) president Dr Azree Shahrel Ahmad Nazri, subjected 150 participants to clinical tests and achieved 94 per cent sensitivity for symptomatic cases among people who had received official positive COVID-19 RT-PCR test results, while 88 per cent are those who had no other symptoms.

“In mass testing with 1,009 participants, it has achieved 94% accuracy, 96% sensitivity and 94% specificity.

“This application, which was named as AI-Screen, is being evaluated by National Technology & Innovation Sandbox (NTIS), Ministry of Science, Technology & Innovation (MOSTI),” Bernama quoted Serba Dinamil Holdings Berhad saying in a statement.

According to the statement further, the AI could practically be used for daily screening of students, workers, and the public, in pool testing exercises.

The company's managing directot and chief executive officer Mohd Abdul Karim Abdullah said that the AI-Screen is a smartphone application which is capable of recording ad sending three cough sounds to an AI engine, which then churns out a result within two months. The application is deployable anytime, anywhere, and by anyone.

However, Abdul Karim said that the result should not be taken as a final confirmation by users, as one has to conduct a RT-PCR test at a government or private health facility for further verification.

As such, the AI-Screen application is not meant to replace or compete with existing medical grade testing, but merely complement them, he added.

Abdul Karim said the application could become accessible to the public by the end of this year, as a pre-screening method for COVID-19.


Source: Bernama
Photo source: Astro Awani