Vuno said that it has proved the efficacy of its artificial intelligence (AI)-based cardiac arrest prediction software, VUNO Med–DeepCARS.

Vuno has published a study confirming the efficacy of its AI-based cardiac arrest prediction software, VUNO Med–DeepCARS.
Vuno has published a study confirming the efficacy of its AI-based cardiac arrest prediction software, VUNO Med–DeepCARS.

It published the study results in Resuscitation, an internationally renowned emergency medicine journal.

The company expects that the study, which verified the clinical effectiveness of the device based on various evaluation indicators in various medical environments, will help its device become an innovative screening tool for cardiac arrest patients in medical fields worldwide.

The study aims to verify whether VUNO Med–DeepCARS consistently and effectively predicts cardiac arrests among patients in various medical settings.

Vuno's biosignal research team used data from 173,368 adult patients admitted for 12 months at five midsize and large medical institutions with different medical environments, such as the size, location, and the presence of a rapid response system.

Based on the data, the research team retrospectively verified three evaluation indicators -- prediction accuracy, false alarm rate, and early prediction power -- compared to the modified early warning score (MEWS).

VUNO Med–DeepCARS proved excellent performance in all evaluation indicators. In external performance verification, the solution's in-hospital cardiac arrest prediction accuracy was 15.3 percent higher than that of MEWS.

Also, when the predicted performance evaluation values such as specificity and number of alarms were the same, the sensitivity was 63.2 percent higher than that of MEWS. The average number of warnings for the same sensitivity decreased by 44.2 percent to MEWS, confirming a low false alarm rate.

The study further confirmed the excellent early predictive power of the company device by comparing the number of patients with in-hospital cardiac arrest predicted at a specific time point before cardiac arrest.

For example, VUNO Med–DeepCARS detected more than twice as many patients with in-hospital cardiac arrest as MEWS 20 hours before the onset of cardiac arrest. The number of predicted cardiac arrest patients in all-time points was higher than that of MEWS.

"This study demonstrated that VUNO Med–DeepCARS is an innovative solution for effectively predicting cardiac arrest in inpatients in a variety of medical settings," Vuno Chairman Lee Ye-ha said. "At the same time, the company also demonstrated the capabilities of Vuno's biosignal research by continuously published research results in world-renowned academic journals."

The company will continue introducing bio-signal-based AI solutions that contribute to saving patients' lives, including VUNO Med–DeepCARS. The company expects to receive product approval within this year, Lee added.

VUNO Med–DeepCARS provides predictive information on the risk of cardiac arrest within the next 24 hours based on five vital signs – diastolic and systolic blood pressure, pulse, respiration, and body temperature -- collected from electronic medical records (EMR) of hospitalized patients in general wards.

Copyright © KBR Unauthorized reproduction, redistribution prohibited