Algorithms to analyze data from intensive care units

Computer-supported systems and Artificial Intelligence (AI) could be used to help classify and evaluate the flood of data from intensive care units. To achieve this, the university of Zurich (UZH) has launched the ICU Cockpit project. In terms of partners, the university has successfully brought on board the Swiss Federal Institute of Technology in Zurich (ETH), IBM Research Rüschlikon and the company Supercomputing Systems. 

“Our aim is to ease the burden on specialists and at the same time improve the accuracy of prognoses”, comments Emanuela Keller, UZH Professor for neurological intensive care medicine and lead physician of the neurosurgical intensive care unit at the University Hospital Zurich in a report published by UZH. According to this report, 700 alarms per patient per day are still being triggered by biosensors. The majority of these are false positives. In addition, every patient reacts differently to therapies. Therefore, in addition to the sensor data, it is important to take into consideration as many parameters as possible, such as diagnoses, clinical data, laboratory values, genetic background and the lifestyle of the individual patient in question.

All of this should benefit what is known as precision medicine. If digitization and machine learning could sort through the millions of units of data that are generated every second, the result could be saved lives. “Our aim is to identify correlations between different laboratory values and the many signals from the biosensors that indicate an impending multi-organ failure”, according to Keller. In addition, it should be possible to predict the severity of a COVID-19 infection at an early stage for every patient by leveraging the benefits of AI.