Emergency authorization granted to COVID-19 ICU prediction software

Dive Brief:

FDA has granted emergency use authorization to software that aims to predict whether a COVID-19 patient will develop dangerously low blood pressure or respiratory failure.
The EUA, issued Tuesday, covers a CLEW Medical system that analyzes data such as vital signs for evidence the condition of a COVID-19 patient will deteriorate.
In a retrospective study, 5.5% of patients flagged as being at risk of respiratory failure by the system went on to need intubation. The technology, called the CLEWICU System, was better at correctly identifying people at risk of low blood pressure.

Dive Insight:

Hospitals dealing with large numbers of COVID-19 patients face the challenge of identifying high-risk individuals at a time when their resources are stretched thin. Data generated through the continuous monitoring of ICU patients could contain clues about which patients are likely to deteriorate, but the sheer amount of information means healthcare professionals may base decisions on snapshots of the available evidence.

In theory, computers are better suited to the task of monitoring streams of patient data and spotting patterns that indicate an individual is at elevated risk of negative outcomes. That thinking underpins the CLEWICU System.

The system sifts through vital sign data, nursing assessments, lab results, details of the medications a patient is taking and other sources of information. When the system spots a trend that suggests a patient is at risk of respiratory failure or hemodynamic instability, it alerts the person caring for them. CLEW Medical contends the advanced notifications could enable healthcare professionals to act to stop respiratory failure or hemodynamic instability from happening, or prepare for when it does.

Evidence of the effectiveness of the technology comes from a retrospective multi-center validation study that CLEW Medical presented in its instructions for use. The positive predictive value, which shows the probability that an alert is correct, for the respiratory failure application of the system was 5.5%. The system typically notified physicians of true positives 3.4 hours ahead of the event.

When used to predict hemodynamic instability, the CLEWICU System had a positive predictive value of 19%. The median lead-time for true positive alerts was close to the respiratory failure application, clocking in at 3.5 hours.

Those results come from a retrospective study. The system is undergoing testing at UMass Memorial Medical Center in Massachusetts and WakeMed in North Carolina, and is being deployed at sites in Israel, where CLEW Medical is based. The company has filed for 510(k) clearance from FDA but, in the meantime, can market its system under the EUA.

The EUA is part of a series of regulatory okays for software products designed to inform the care of COVID-19 patients. Earlier this month, Eko Devices and Caption Health received FDA clearance for software to help analyze and capture data relevant to the risk of cardiac complications.