The TRUStroke project proposes a new platform based on artificial intelligence, which respects privacy, to help physicians, patients and caregivers in the management of ischaemic stroke, both acute and chronic. The platform integrates clinical data and information provided by patients to provide a reliable assessment of disease progression and risk, enabling personalised stroke management and the sharing of best practice between hospitals.

TRUStroke will address the following risks

1. Clinical severity at discharge.
2. Clinical deterioration leading to unplanned hospitalisation.
3. Poor mobility, incomplete recovery and long-term adverse clinical outcomes
4. Recurrent stroke

To achieve these goals, a federated learning infrastructure will enable multiple clinical sites to build reliable AI-based predictive models using stroke data without compromising privacy and implementing best security protocols.

Standardised clinical data (FAIR) from major European hospitals and data from a remote home care system will be used to

Train and validate reliable AI models for stroke prediction.

Personalise cardiovascular risk assessment, treatment adherence and communication with healthcare professionals.

The TRUStroke platform will be designed to be reliable and compliant with recognised guidelines for creating transparent, explainable, robust, accountable, accurate and secure AI assets and systems.

User experience studies will be conducted to improve the usability of the platform and communication with end users. Finally, a demonstration clinical study conducted by world-class stroke centres will ensure the highest level of reliability of the TRUStroke platform.

Project coordinator: Vall d’Hebron Institute of Research (VHIR)