ICU Cockpit
A dashboard for the Neurosurgical intensive care unit of the University Hospital Zurich
In today’s intensive care units (ICUs), biosensors collect up to 100 GB of data per patient every day. These vast amounts of data are only valuable if they lead to better patient outcomes.
- How can medical staff gain insights from this data?
- How can risk constellations be identified as early as possible?
- How can therapeutic recommendations be adapted?
In the context of multiple research projects, our team is developing the "ICU Cockpit" user interface, a dashboard for the Neurosurgical intensive care unit of the University Hospital Zurich.
The dashboard visualizes patient data and makes algorithmic predictions accessible to ICU health professionals. It supports medical decisions and thereby improves patient safety.
Dashboard of the ICU-Cockpit showing an overview of patients and their key metrics. Real-time data of various signals can be viewed for each patient.
In spring 2020, during the early phase of the COVID-19 pandemic, our team developed a "COVID-19 Dashboard" to display patient isolation requirements. The dashboard visualizes the arrangement of beds in the ICU, and displays the most recent SARS-CoV-2 testing/vaccination status for each patient. In addition, the user interface displays vital parameters such as heart rate or ECG, and allows for remote patient video surveillance.
”COVID-19 Dashboard” at the front desk of the ICU (red arrow) providing the state of SARS-CoV-2 testing and video monitoring for all patients.
Among subarachnoid hemorrhage (SAH) patients who survive a ruptured aneurysm, delayed cerebral ischemia (DCI) is the most important cause of poor neurological outcome or even death.
Based on routine biomarkers and high-resolution data from the ICU Cockpit, machine learning models can provide accurate predictions of future DCI events.
To support clinical practice in the ICU, algorithmic predictions must be accessible through a human-centered interface.
Early prototype of DCI prediction, integrated into existing ICU Cockpit user interface.
Current SARS-CoV-2 PCR tests have a rather high false-negative rate even after symptom onset. This is problematic for effectively ruling out virus transmission among patients and medical staff. Additionally, COVID-19 patients could greatly benefit from an early warning of worsening disease progression.
Based on routine biomarkers and high-resolution data from the ICU Cockpit, machine learning models can provide a more accurate diagnosis as well as a live prognosis of the course of disease.
Added to the ICU Cockpit “COVID-19 Dashboard”, they will support the medical staff in the ICUs by faster and more reliable recognition of SARS-CoV-2 infected patients and benefit severely ill patients through a more personalized therapy.