Using machine learning, this test detects SARS-CoV-2 in five minutes

The Department of Physics at the University of Oxford in the UK has developed a new type of test able to detect the presence of SARS-CoV-2 with great precision and in five minutes thanks to a machine learning system.

A test for mass screening

The team of researchers says this new Covid-19 test would be of great help in quickly screening people in places open to the public, such as concert halls, businesses and airports.

Our method quickly detects intact viral particles; which means the test is simple, extremely fast and cost effective“said researcher Achillefs Kapanidis, who took part in the work.

The method begins with a sample taken from the patient’s throat. Using a microscope, the sample is analyzed to detect viral particles, markers of SARS-CoV-2 infection. These images are analyzed by software based on a machine learning system capable of “quickly and automatically” identifying the virus present in the sample, exploiting the fact that each virus has a distinct size and shape.

A much faster test

If this new type of test is faster, it is because it does not require amplifying the RNA molecules in the samples to isolate them from other components, unlike conventional virological tests (PCR). According to scientists, their test would have a success rate of 98%. This figure should however be put into perspective because the manipulations only took place under optimal conditions in a research laboratory.

To be precise, the publication is a preprint, that is to say it has not yet been approved by the editorial board of a scientific journal. It is therefore advisable to remain cautious about the future marketing of this test, scheduled for 2021 according to scientists.

In the meantime, the team of researchers are working with Oxford University Innovation and a consulting firm to create a spin-out. They are also seeking funding to accelerate the transformation of the test into a fully integrated device that will be deployed as a real-time diagnostic platform, capable of detecting multiple virus threats.

Leave a Comment