Smarter decision-making for intensive care units
During the first waves of the COVID-19 pandemic, intensive care unit (ICU) staff were faced with a new, highly contagious disease for which there was no vaccine. In severely hit countries, more than 10 % of infected patients were requiring intensive care, putting immense pressure on resources and placing medical staff at risk. “Here in Frankfurt, my team and I were treating over 700 severely compromised ICU COVID-19 patients at one time,” says ENVISION project coordinator Kai Zacharowski from Goethe University Frankfurt in Germany. “While we knew how to ventilate and carry out other ICU treatments, we lacked valid, proven treatment options for this complex disease. This made it extremely difficult to always make decisions that were in the best interests of patients.”
Making effective medical treatment decisions
The ENVISION project, which began in December 2020, sought to address this health challenge, by developing a data-driven tool to help ICU medical staff in the treatment of patients. For this purpose, data was collected from a range of sources, including ventilators and medical imaging devices, and analysed by artificial intelligence (AI). A total of 13 hospitals across Europe were involved. “A small start-up company agreed to be on board,” adds Zacharowski. “A software programme was developed and run in ICUs, enabling all physiological, treatment and medication data to be gathered.” An existing digital app, called Sandman.MD was further developed to become the Sandman.IC. This was designed to be an intelligent decision-support system for staff, helping in the monitoring, prediction and treatment of COVID-19 patients in ICUs. The idea was that the tool might be able to quickly suggest possible novel treatment options that might not otherwise be identified by busy medical staff. The project team also hoped that this gathered information would increase our understanding of the yet unknown course of this disease, and support other medical efforts in finding new therapies.
Gathering and integrating medical data
A key project success was the ability to gather and integrate medical data from across Europe, while abiding by strict data protection laws. “We achieved this by leaving hospital data with the hospitals,” explains Zacharowski. “Data was analysed in hospitals, and anonymised results then transferred.” The introduction of the app in clinical settings was also successful. Sandman.IC was integrated into an AI-driven data analytics suite with predictive modelling tools, and enhanced with a smart alert functionality to make it easier to use. The tool has since been validated, and is ready for market approval.
Tackling future infectious disease outbreaks
The potential impact of the ENVISION project goes beyond immediate results, however. The knowledge gained from analysing COVID-19 patient data, and the development of an AI tool that supports clinical decision-making, could be transferred and applied to future infectious disease outbreaks and other critical care scenarios. This will not only improve patient care during health crises, but also help ensure that ICUs are run as efficiently as possible. A key lesson from the COVID-19 pandemic was the importance of preparedness. “We succeeded in connecting European countries to increase our collective knowledge on this,” notes Zacharowski. “In the future, we will be looking more at how we receive and analyse data using AI.”
Keywords
ENVISION, AI, COVID-19, disease, medical, ICU, contagious, vaccine
