After a major surgery, there is an approximately 30% risk for serious adverse events during the first 30 postoperative days, and postoperative complications are ranked as the third most likely cause of death worldwide.

“We wanted to develop a system that offered the anesthesiologist’s level of continuous monitoring at the general ward, but tailored to the specific conditions there. This included a focus on detecting true relevant alerts and reducing irrelevant false alerts.”

“These numbers are not due to complications occurring in the OR, PACU, or ICU, but mainly in the general ward, where nurses are responsible for high patient volumes and infrequent intermittent manual vital sign monitoring result in late detection of complications and failure to rescue,” said Eske Kvanner Aasvang, MD, DMSci, Professor and Chief Consultant, Department of Anesthesiology, Copenhagen University Hospital-Rigshospitalet, Copenhagen, Denmark.

Despite decades of use, international standards for track-and-trigger systems, such as the National Early Warning Score (NEWS), have not been proven to have any effect on morbidity or mortality.

Recognizing this, Dr. Aasvang and his colleague Christian S. Meyhoff, MD, PhD, Professor and Chief Consultant, Department of Anesthesia and Intensive Care, Copenhagen University Hospital – Bispebjerg and Frederiksberg, Copenhagen, Denmark, set out to develop a system using data from vital sign sensors to identify patients at risk of complications after surgery, medical admissions, or at home.

The Wireless Assessment of Respiratory and Circulatory Distress-Clinical Support System (WARD-CSS) is intended to alert health care personnel to oncoming or actual vital sign deviations and clinical complications at the general ward or in the home.

“We wanted to develop a system that offered the anesthesiologist’s level of continuous monitoring at the general ward, but tailored to the specific conditions there,” said Dr. Meyhoff. “This included a focus on detecting true relevant alerts and reducing irrelevant false alerts. Over-alerting is a major problem with existing continuous monitoring equipment. It is estimated that, without AI modification, 90% of alerts are considered irrelevant, which results in alert fatigue, and ultimately the real alert is often ignored, too.”

Drs. Aasvang and Meyhoff used commercially available wireless vital sign sensors (e.g., ECG patch, pulse oximeter, blood pressure device) to collect vital signs in real time, relay it to a cloud server that hosts their artificial intelligence (AI) algorithms, and interpret the data. The vital sign data is then transmitted to the nurse’s smartphone, along with an overview of the current condition developed in collaboration with the nursing staff, and alerts when criteria for clinically relevant deviations are met.

What began as a collaboration between two major hospitals in Copenhagen – Rigshospitalet and Bispebjerg Hospital – and the Technological University of Denmark has since grown into a national project that includes several other hospitals and research groups.

Pilot studies as well as a series of prospective surgical and medical studies with more than 2,000 patients have been completed. The research team is now finalizing the first of two randomized clinical trials in 700 patients who will be allocated to standard of care or the WARD-CSS system.

The next phases include more data-driven alerts based upon their database, including earlier detection and unique AI alerts that increase the accuracy regarding true alerts, explained Drs. Aasvang and Meyhoff.

Current research has shown that WARD-CSS significantly reduces alerts compared to simple continuous vital sign monitoring by 80% and detects signs of complications several hours before the clinical staff, according to yet-to-be published data. To date, Drs. Aasvang and Meyhoff have not observed any signs of device-related side effects from using the WARD-CSS.

“Improving post-ICU/OR/PACU care via a system such as the WARD-CSS is crucial to perioperative medicine. WARD-CSS allows us to identify patients in need and gain an understanding of physiology at a detail we have not had before,” they said.

“This will, without question, result in new insights that will challenge our concepts of what a vital sign and deviation is,” they continued. “For example, our two prospective studies have shown that long cumulative durations of vital sign deviations also occur in patients without complications, but AI allows us to identify patterns that are more accurate for finding patients with an oncoming complication.”

“Overalerting is a major problem with existing continuous monitoring equipment. It is estimated that, without AI modification, 90% of alerts are considered irrelevant, which results in alert fatigue, and ultimately the real alert is often ignored, too.”

A commercial version of the system has been submitted for Conformitè Europëenne (CE) Mark and FDA approval, which is expected in 2023, and the researchers plan to conduct international pilot studies in Scandinavia, Europe, and the United States within the next year.

Learn more about WARD (asamonitor.pub/3hv57K3).