Recovering in peace

Beep, flash, buzz … the neurosurgical intensive care unit is a hectic place, due in part to the many alarms sounding from various technical devices. Artificial intelligence could help reduce the number of false alarms – as more peace and quiet would be good for the health of both patients and staff.

I would like more information →

The 12 beds in the neurosurgical intensive care unit at University Hospital Zurich are reserved for people with potentially life-threatening conditions – for example, patients who have suffered a stroke or a brain hemorrhage. “It’s a hectic place. This also takes its toll on the staff,” says Professor Emanuela Keller, Head Physician at the Intensive Care Unit.

The commotion is caused in part by the various technical devices, which are used to monitor the patients’ brain activity, breathing and heart function. The equipment takes up to 1,000 readings per second. If one of the readings is abnormal, the device sounds an alarm – this happens about 50 times per patient, per day. And false alarms sound about another 250 times per patient, per day.

But, thanks to the ICU Cockpit project, this is now set to change. Professor Keller’s team is hoping that artificial intelligence can help them consolidate the huge data volumes, using algorithms to organize the many readings into categories. This would make it easier to identify a real emergency – for example, when certain respiratory, cardiac and cerebral readings all occur simultaneously. In the future, situations like this would also be clearly displayed on a central device, significantly reducing the number of false alarms.

The project has been made possible thanks in part to a donation to the USZ Foundation from entrepreneur Hans-Peter Wild. Keller’s team has already developed algorithms that help predict the risk of further brain damage. The algorithms used to prevent false alarms are scheduled to be introduced in the next step. Keller is excited about the development: “I am confident that this will reduce risks for patients and considerably improve working conditions for the staff.” The findings could also be of interest to other intensive care units.

100% financed

Institute for Biomedical Engineering, ETH Zurich and University of Zurich
Integrated Systems Laboratory, ETH Zurich
Mobile Health Systems Laboratory, ETH Zurich
Institute for Medical Engineering and Medical Informatics, FHNW
IBM Research
Supercomputing Systems AG

Dr. Hans-Peter Wild