FDA greenlights ICU AI for predicting fatal COVID-19 complications

A patient in a hospital bed on a ventilator
According to CLEW Medical’s validation studies, its system predicted oncoming respiratory failure or hemodynamic instability by an average of three to four hours. (Getty/sudok1)

The FDA granted an emergency authorization to CLEW Medical’s remote data monitoring system to help predict and identify COVID-19 patients under intensive care who are most at risk for respiratory failure or insufficient blood flow.

The company’s CLEWICU artificial intelligence software ingests a large range of patient data by linking to electronic medical records, connected medical devices and vital sign monitors, with no need for additional data entry by ICU staff. 

Sponsored by GenScript

Accelerate Biologics, Gene and Cell Therapy Product Development partnering with GenScript ProBio

GenScript ProBio is the bio-pharmaceutical CDMO segment of the world’s leading biotech company GenScript, proactively providing end-to-end service from drug discovery to commercialization with professional solutions and efficient processes to accelerate drug development for customers.

In addition to alerting providers to a patient’s potentially deteriorating condition as a diagnostic aid, the system’s real-time risk classification model also predicts which patients may not require any interventions within an eight-hour period. 

RELATED: FDA looks to expand remote monitoring to free up hospital space

The FDA also said the standalone CLEWICU software, based within the hospital or in the cloud, could help reduce contact between healthcare workers and COVID-19 patients through remote monitoring.

“By identifying high risk patients, prioritizing treatment based on patient acuity, and reducing the false alarm rate this likely provides better care for patients and reduces the strain on the ICU personnel,” the agency said in its letter granting the Emergency Use Authorization. 

According to CLEW’s previous validation studies, the system was able to provide true positive alerts predicting oncoming respiratory failure or hemodynamic instability by a median of three-and-a-half hours. 

The company previously submitted an application for a full 510(k) FDA clearance while the system undergoes testing at tele-ICU facilities at UMass Memorial Medical Center and the WakeMed health system in Raleigh, North Carolina.

Suggested Articles

Researchers at DeepMind believe they’ve cracked the code with a computer model that can translate an amino acid chain into a 3D protein structure.

A computer model developed by MGH researchers identified signs in breast cancer images that could predict a woman’s chances of developing the disease.

U.K.-based alpha-synuclein biotech Wren will partner with Japanese pharma Eisai to seek out new ways of targeting neuro disorders.