AI app to read lateral flow tests for COVID-19 helps reduce false results

An AI app to read lateral flow tests for COVID-19 has helped reduce false results in a new trial today.

Posted in Medicine Cell ReportsA team of researchers from the University of Birmingham, Durham University and Oxford University has tested whether a machine learning algorithm can improve the accuracy of results from antigen lateral flow devices for COVID-19.

The LFD AI Consortium team worked at the testing centers with the help of the UK’s Health Security Agency and with healthcare workers who conducted a self-test to pilot the AI ​​application. More than 100,000 images were submitted as part of the study, and the team found that the algorithm was able to increase the sensitivity of the results, determining between a true positive and a false negative, from 92% to 97.6% accuracy.

The widespread use of antigen lateral flow devices has been an important moment not only during the pandemic, but it has also provided diagnostic tests for many people in the community. One of the drawbacks of the LFD test for Covid, pregnancy and any other use in the future is the ‘faint line’ question – as we can’t tell if it’s positive or not.

The study looked at the feasibility of using machine learning to take the guesswork out of fading line tests, and we’re pleased to see that the app has seen an increase in the sensitivity of the tests, reducing the number of false negatives. The promise of this type of technology can be used in a lot of applications, to reduce uncertainty about test results and to provide critical support for visually impaired people.”

Professor Andrew Bigs, Professor of Cancer Genetics and Surgery, University of Birmingham and lead author of the study

Professor Camila Caillado, Professor of Statistics at Durham University and chief statistician on the project, said:

“The increase in sensitivity and overall accuracy is important and shows the potential of this application by reducing the number of false negatives and future hits. Importantly, the method can also be easily adapted to the evaluation of other digital readout devices of the lateral flow type devices.”

source:

Journal reference:

LFD AI Federation. , (2022) Machine learning to determine lateral flow device results for testing for SARS-CoV-2 infection in asymptomatic populations. Medicine Cell Reports. doi.org/10.1016/j.xcrm.2022.100784.

Originally published at San Jose News Bulletin