Stroke is a condition where poor blood flow in the brain results in cell death. This can lead to a part of the brain not functioning properly, with signs and symptoms appearing soon after the stroke has occurred. Time is critical in acute stroke care: within a very small time frame of emergency treatment, health professionals need to identify the type of stroke and severity, decide upon the treatment, transport the patient to the adequate care centre, and perform the required intervention. The acute treatment of stroke generates and requires a large amount of data that need to be shared between the health professionals along the whole process. Such data also represent valuable sources of evidence for medical research afterwards. Cloud infrastructures offer an attractive solution for sharing such data, but the sensitive character of the data, in general, raises privacy and safety issues.
ASCLEPIOS in the Stroke Acute Care
The vision of ASCLEPIOS is to develop security enablers that will safeguard the privacy of patients when their data are shared or processed on cloud infrastructures. These solutions can potentially be exploited to enable and facilitate data sharing for stroke treatment and research. The Amsterdam University Medical Centres, University of Amsterdam (AMC) is a leading player in stroke treatment in the Netherlands and a partner of the ASCLEPIOS consortium. AMC will implement a demonstrator of the ASCLEPIOS solutions that will enable secure data sharing during acute stroke treatment and for research purposes.
The ASCLEPIOS stroke demonstrator will explore two scenarios: hyper-acute care and construction of predictive models.
Case 1 – Hyper-Acute Care
The use of electronic medical records (EMR) improves the overall quality of care that a patient receives because their use can lead to a substantial reduction of unnecessary investigations and improve communication between the healthcare professionals involved in the treatment. Therefore, the availability of EMR, especially when a patient is under an emergency situation, is of paramount importance.
The hyper-acute care demonstrator will implement a cloud-based EMR that is secure and preserves patient privacy. The demonstrator will use modern encryption schemes to dynamically grant and revoke authorisation to access patient information during the hyper-acute phase. This will illustrate how to enhance secure and GDPR-compliant patient information sharing during the acute phase of a patient’s journey from the place of stroke onset to the centre of care.
Case 2 – Construction of Prediction Models
A stroke patient stays hospitalised for a few days after treatment due to potential risk of complications. For example, delayed cerebral ischemia (DCI) is one of the most severe complications in patients that had a hemorrhagic stroke. Currently, it is very difficult to predict DCI, so risk assessment takes a conservative approach, potentially extending hospitalisation longer than necessary. A variety of machine learning methods could be used to develop DCI prediction models that combine clinical and imaging data. Obtaining large amounts of data, as required for learning the models, has been challenging. Not only high-quality data are scarce, but also data controllers are reluctant to share. Another challenge is organising the computing infrastructure to learn the models, which requires HPC resources when imaging data is involved. Although the data are anonymised, there are still restrictions about transferring them to infrastructures outside the hospital intranet. Moreover, revocation of authorisation when access to the data is no longer necessary, is by itself another thorn in health data sharing. These difficulties altogether have discouraged data controllers from sharing their data. ASCLEPIOS will exploit privacy-preserving techniques and perform analytics on cloud-based EMR data for predictive modelling in stroke care research.
The stroke demonstrators built in ASCLEPIOS will illustrate novel manners for sharing and processing data on clouds in a secure and GDPR-compliant manner. These demonstrators will pave the way for improvements in healthcare systems towards higher data availability without compromising privacy and confidentiality.