AI Based Device for Covid Testing
Researchers at the University of the West of Scotland (UWS) devised a testing platform that can detect the SARS-CoV-2 virus much faster than a PCR test, which can take up to two hours. The test involves using artificial intelligence (AI)-enabled X-Rays to reliably diagnose COVID-19.
The process could potentially be used to reduce pressure on overburdened hospitals and will come in aid in places where PCR tests are not commonly available.
The method uses X-Ray equipment to compare scans to a database of roughly 3000 images from COVID-19 patients, healthy people, and people with viral pneumonia. Thereafter, a diagnosis is made using an AI method called a deep convolutional neural network, which is normally used to analyze visual imagery.
During an exhaustive testing phase, the technique has proven to be more than 98 percent accurate, according to a study in the journal Sensors.
Professor Naeem Ramzan of UWS, who led the study, said, “There has long been a need for a quick and reliable instrument that can detect COVID-19, and this is becoming even more true with the upsurge of the Omicron variant. Several nations are unable to conduct significant numbers of COVID-19 testing due to a lack of diagnostic instruments, but this technique uses readily available technology to detect the virus fast”.
Since COVID-19 indications are not 100 percent obvious in X-Rays during the primary infection stages the technology cannot completely replace PCR tests as of now, according to the researchers.
The researchers now intend to expand the study by including a large dataset of X-Ray images obtained by other brands of X-Ray equipment in order to assess the approach’s applicability in a clinical context.
Machine Learning Based Tool
A German research group has developed an artificial intelligence program that can anticipate how a covid patient would react after being admitted to the hospital-based on a blood sample.
The machine learning technique can reliably forecast whether a person will live or die based on protein levels in their blood.
It is designed to assist health care providers in determining if a patient needs intensive care or can combat the infection on their own. Before making a judgment, the prediction model looks for 14 different protein levels in a patient.
The program with machine learning was conducted on 50 patients at Charité University Hospital in Germany, according to the researchers. According to reports, 15 of the 50 patients died while researchers looked for protein levels and constructed the tool using data from the patients.
After the initial experiment, researchers assessed 24 coronavirus patients who were treated for the same and found that 19 had survived and five had died.
The AI technology, according to the researchers, correctly anticipated the five deaths.
The sample size was small, with 50 patients and 24 other people who were asymptomatic or potential carriers during open trials, according to the researchers.
The latest discoveries come as the world is fighting with the Omicron wave, with hospitals in numerous places becoming overburdened. Several health professionals have been infected as a result of the Omicron wave, which has made it difficult for hospitals to function.