Algorithm Predicted of Lung Damage by Coronavirus Developed
The tool detects acute respiratory syndrome early in the infection
A group of researchers from the United States and China reported Monday that they developed an artificial intelligence tool to accurately predict which newly infected covid-19 patients can develop severe lung damage.
It is an algorithm that can assist physicians in making decisions about where to prioritize resources in systems that are at the limit of the pandemic, explained Megan Coffee, a doctor who is an academic at the Grossman School of Medicine in New York.
Cofee is a co-author of the study published in the journal Computers, Materials & Continua.
This tool revealed several surprising indicators that are indications that predict which patients will develop a condition known as an acute respiratory syndrome (Ards), which is a severe covid-19 complication that fills the lungs with fluid and is fatal in 50 years. Per cent of cases that develop this pathology.
The team used a machine learning algorithm by inputting data from 53 coronavirus patients at two hospitals in Wenzhou, China, and found that changes in three markers were the most reliable clues: levels of the liver enzyme alanine aminotransferase (ALT), body aches and haemoglobin levels.
Using that information, combined with other factors, the tool was able to predict the risk of developing respiratory syndrome with 80 per cent accuracy.
In contrast, markers that were considered distinctive of the virus, such as a pattern in images taken of the lungs called “ground-glass opacity,” fever, and robust immune response, did not help predict which patients started with mild symptoms, they later developed lung pathologies.