TRUSTWORTHY AI FOR IMPROVEMENT OF STROKE OUTCOMES

TRUSTroke proposes a novel trustworthy by design and privacy-preserving AI-based platform to assist clinicians, patients and caregivers in the management of acute and chronic phases of ischemic stroke, based on the integration of clinical and patient-reported data, outcomes and experience for a trustworthy assessment of disease progression and risks. enabling more personalised and effective management of stroke, as well as providing inter-hospital benchmarking and sharing best practices. Specifically, TRUSTroke will address to risks of: i) clinical severity at discharge ii) clinical worsening leading to unplanned hospital readmissions, iii) poor mobility, incomplete recovery and unfavourable clinical long-term outcomes; iv) stroke recurrence.To this purpose, a Federated Learning infrastructure will enable multiple clinical sites to build several trustworthy AI based predictive models by leveraging stroke data without compromising privacy and implementing best-in-class security and privacy protocols. FAIRified clinical data from leading European hospitals and outpatient monitored data from a remote home-care system, will be used (i) to train and validate trustworthy AI models for stroke prediction; (ii) to personalise patients´ assessment of cardiovascular risk factors, treatment compliance and communication with healthcare professionals. TRUSTroke platform will be trustworthy by design since it will be compliant with the recognized guidelines for building FAIR resources and trustworthy AI systems, including the need for transparency, explainability, robustness, accountability, accuracy and security of the learned AI models. A series of User Experience studies will be performed to increase the usability of the platform and improve the communication to the end-users. A final proof of concept clinical study, conducted by world class stroke centres, will ensure the highest level of trustworthiness of TRUSTroke.