Study published in the international journal Acute and Critical Care, issued by the Korean Society of Critical Care Medicine

Prospective external validation using data from over 6,000 hospitalized patients—67% reduction in false alarms, 42% improvement in prediction accuracy

AITRICS (CEO Kwang joon Kim), a company specializing in artificial intelligence(AI) technology announced on the 11th that a prospective external validation study proving the predictive performance of its AI-based patient deterioration prediction solution, AITRICS-VC (VitalCare), has been published in Acute and Critical Care (ACC), an international journal issued by the Korean Society of Critical Care Medicine.

This research was conducted in collaboration with Professor Ho-chan Cho's team from the Division of Endocrinology and Metabolism at Keimyung University Dongsan Hospital. AITRICS analyzed data from 6,039 adult patients hospitalized in general wards and evaluated the predictive performance of VitalCare compared to existing scoring systems such as the National Early Warning Score (NEWS) and the Modified Early Warning Score (MEWS).

The results showed that VitalCare outperformed NEWS and MEWS in predicting major adverse events, such as death, cardiac arrest, and unexpected ICU transfers. Under the same specificity conditions, VitalCare demonstrated approximately 42% higher sensitivity, enabling more accurate prediction of acute events in general ward patients.

In addition, when comparing under the same sensitivity conditions, VitalCare showed significantly improved specificity, reducing false alarms by about 67% compared to NEWS and MEWS.

Originally developed based on data from a single institution, VitalCare demonstrated strong predictive performance in a different hospital environment through this study, proving its applicability in diverse clinical settings.

This research is particularly meaningful as it not only validated the predictive performance of VitalCare but also confirmed its practical utility in real-world clinical settings. The consistent performance across different hospital environments provides critical evidence for the solution’s generalizability and scalability.

Taeyong Sim, CMO of AITRICS, commented, “This study goes beyond proving VitalCare’s performance—it shows that data-driven medical AI can function as a trusted tool in real clinical settings. Given the diversity in patient populations and operational environments across hospitals, consistency and adaptability of algorithms are fundamental competitive advantages for medical AI. We will continue to build clinical evidence across various settings to contribute meaningfully to patient safety.”