Can AI help diagnose COVID-19? After testing waters in China, tech company eyes US hospitals
Artificial intelligence (AI) is believed to have tasted success in diagnosing COVID-19 in some hospitals in China and Russia. The makers of the technology are now eyeing the US and Europe.
Powering the AI is a company named RADLogics. To get its technology up and running in the US, the company has applied for the emergency use authorization from the US Food and Drug Administration (FDA). Under this category, the FDA approves new drugs or diagnostic tests during a declared emergency such as the COVID-19.
Though some doctors are not convinced, some experts believe AI can boost the testing capacity of countries. "AI is of particular value in places where disease prevalence is high and availability of testing kits is limited," Adam Bernheim from Icahn School of Medicine at Mount Sinai, told MEA WorldWide (MEAWW).
If the FDA approves RADLogics's AI, doctors can use it to analyze lung or CT scans of suspected patients, saving their time. Normally, doctors use lung scans to look for clues that indicate a patient is suffering from pneumonia. Experts hope that AI can replace doctors and save their time. Additionally, the technology can help track a patient's response to a drug.
RADLogics is not alone. Chinese tech giant Alibaba developed an AI technology capable of detecting COVID-19 in 20 seconds with 96% accuracy. "Our [AI] system is now being used by 26 hospitals in China and has helped diagnose over 30,000 cases," AI expert Xu Minfeng told Alizila.
What is AI and why it is important?
AI has generated a lot of interest. It means using computers to understand and mimic human intelligence. The technology has come a long way since it was first use in 1947. Most recently, scientists have developed an AI capable of turning thoughts into words.
The technology has assisted doctors by spotting cancers. One study found that AI outperformed humans in detecting diseases from scanned images.
Dr Bernheim worked on a study evaluating AI in detecting COVID-19 from lung scan images. He said: "The early study shows that there may be a role for AI deployment in the setting of an acute pandemic and helps lay the groundwork for future investigation."
Concerns over using AI
For all its benefits, AI is not perfect. It could make errors. To make AI reliable, experts train it using patient samples. When the AI is exposed to enough patient scans, it identifies patterns that are specific to the lung damage brought about by COVID-19. But experts fear AI could be unreliable due to a limited number of lung scans.
“I’ve looked at the workaround AI and COVID, and I’m not a big fan,” Ella Kazerooni, a radiologist at the University of Michigan Medical Center, told STAT.
She questioned whether AI can differentiate between COVID-19 and other lower respiratory illnesses. She added that AI is trained using images from critically ill patients and may not represent a setting.
What is more, some doctors are hoping to limit the use of CT scans due to the risk of contamination. When an infected person undergoes a CT scan, the virus might contaminate the equipment, potentially exposing others who come into contact with it.
In response, Dr Bernheim said: "Protocols are in place following CT scanning of a known or suspected COVID-19 patient to ensure that risk of transmission to future patients is minimized."