AI-4-NICU - Maximising impact of multidisciplinary research in early diagnosis of neonatal brain injury.

From 19-10-2021 to 18-10-2025

Description

The overarching aim of this COST Action is to improve neurological outcomes for infants with NBI through the provision of enhanced neuro-monitoring technologies in the NICU; at the heart of these AI-technologies are cutting-edge ML algorithms. The interdisciplinary AI4NICU network will create a pan-European partnership that will share learning, resources, and pool data to create curated open-access databases. Coordinated effort, consensus on protocols and best practice, and open data and algorithms will accelerate research in this field. 

Europe is facing major health challenges which need to be addressed to ensure a secure future and a sustainable, competitive economy. The Covid-19 pandemic has placed further pressure on health-care services and the economy. The life-long consequences of NBI can have devastating effects on the health of European citizens and place additional economic strain on health systems. 

The solutions are likely to be beyond the scope and resources of any single country and must be addressed collectively. Hence, there is a need for better coordination of research within Europe, around the world, and between academia and industry. Many researchers are working in the field of ML model development for NBI. As mentioned previously, there is active research in seizure detection, HIE grading, and sleep staging for example [11–15,20,21]. Much of this effort is scattered at a national level, or at best individual European efforts. These fragmented efforts—using different clinical protocols for acquiring and managing data, different techniques for testing algorithm performance, and locked data sets and algorithms—lack reproducibility and therefore utility to others. The lack of synergies impedes progress. 

The rapid and recent development of advanced statistical modelling and ML methods has enormous potential in the data-dense field of neonatal brain research. To maximize the potential of combining the fields of neonatal brain health and ML, several challenges must be overcome. There is currently a lack of optimal understanding and implementation of ML modelling within the NBI field, which can lead to overoptimistic, misleading, or sometimes erroneous results. Few concerted efforts so far have been made, with notably exceptions such as a multi-centre clinical validation of a seizure-detection algorithm [10,13] or the INFANS training network for ECIs, and a limited community currently exists beyond isolated groups. We group several technical challenges as follows. 

Team

• Maarten De Vos, Co-promoter
• Gunnar Naulaers, Coordinator (External)

Financing

Funding: EU Funding - European Funding

Program/Grant Type: EU COST - European Cooperation in Science and Technology