Data sharing for improved decision in stroke acute care by the Amsterdam UMC

Time is critical in acute stroke care: within a very small timeframe 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 among the health professionals throughout the whole process. Such data also represent valuable sources of evidence for medical research afterwards.

At the University of Amsterdam (Amsterdam UMC), we implemented the Stroke Acute Care demonstrator consisting of a system for secure patient information sharing during emergency treatment and its reuse for research purposes. It relied on a unified patient’s EMR stored in a cloud system in encrypted form to improve accessibility and on medical records collection during an emergency.

1.1.1    Architecture of the Stroke Acute Care Demonstrator

The following figure presents the demonstrator’s architecture, which covers the needs of both the stroke acute care and the research use cases.

Figure 1: Architecture overview of the Stroke Acute Care demonstrator. Diagram of the communication between the demonstrator and the ASCLEPIOS framework

All the users interact with ASCLEPIOS components through the web application interface.

  1. To start communicating with the ASCLEPIOS services, the user first needs to be authenticated at the web application server.
  2. The Registration Authority (RA) is responsible for authenticating and authorising the user with an authorisation token.
  3. RA generates the user’s secret, an Attribute-based Encryption (ABE) key, and handles it to the user through the web application.
  4. The web application interface contains the client-side of the Symmetric Searchable Encryption (SSE) and Functional Encryption (FE) services, which are responsible for performing cryptographic functions over the encrypted data.
  5. All communications between these clients and servers go through the Attribute-based Access Control (ABAC) service.
  6. Some ASCLEPIOS components, such as the Keytray and FE services, execute their functions within the trusted boundary of the Trusted Execution Environment (TEE) implemented through a secure and encrypted enclave.

1.1.2    Key results from the demonstrator

Hospitals, including the Amsterdam UMC, are well known for being very strict about security and for being secretive about it. It all happens behind the doors of the ICT department and is experienced by everyone else as “things we are not allowed to do”. ASCLEPIOS was one of the first projects to directly address cybersecurity as a research topic in our organisation, and to initiate discussions about this. The project activities have enabled us to learn and to make people aware of new concepts and opportunities. Let’s take a closer look:

Use Case 1: Stroke Acute Care

A patient’s EMR is encrypted and stored in a cloud system to improve accessibility for reading and adding new data during an emergency. The demonstrator leverages ASCLEPIOS services to secure the data access and sharing among teams, ensuring that a treatment team only has access to the patient’s data under an emergency condition and for the time needed to complete a specific treatment. The medical data are encrypted with a unique key for each patient, and the healthcare professionals can get access to the key only while they are treating this patient.

The developed web interfaces and the supported stroke acute care information flow were demonstrated and discussed with stakeholders from various organisations (emergency call centres, ambulance services and hospitals) in the form of simulations and interviews. The interviewees were very interested in understanding the technology and excited to give feedback. It became evident to us during the interviews with healthcare professionals that they experience – daily – the lack of a properly connected and secure information infrastructure for patient data exchange. Additionally, a large number of non-interoperating systems is used in practice and replacing them with new systems – like the AMC demonstrator – does not seem realistic in the near future. A path that could be explored involves investing in an interoperation layer for secure and trusted data exchange that could bridge legacy systems with the newly developed technology.

The developed prototype served as a means to disseminate a new vision for secure data exchange during emergencies, attracting interest from stakeholders and raising awareness and -hopefully – demand for these types of services.

Figure 2: Example of the patient’s interface for information input about allergies and intolerances

 

Figure 3: Example of the patient’s home interface

 

Figure 4: Example of the call centre employee’s interface for an emergency episode


Use Case 2: Data Reuse for Research

The medical records generated during the emergency treatment of stroke are valuable for medical research aiming at treatment improvement. In particular, saving time during treatment is crucial, so researchers revise the records to find out ways to optimise the process and decisions. In this use case, a research dataset was generated from the medical records generated with the application developed for use case 1. This new dataset was stored on a secure cloud system where the data are partitioned into subsets, each one encrypted with a unique key. After an authorisation process, researchers are given the right to use one or more of the keys and, therefore, they may access and decrypt different parts of the dataset. The implemented solution offers automated fine-grained access control to research data. Additionally, we also exploited Functional Encryption (FE) to further protect yet enable data reuse. Sensitive timestamp features captured during stroke acute care were encrypted using FE, and basic computations were implemented to calculate treatment duration based on the encrypted timestamps. Authorised researchers can run these functions and retrieve their outcomes (e.g. time passed between two timestamps) without accessing the timestamps values. Web interfaces implement functions for different user roles: the Principal Investigator (PI), who manages who has access to which particular dataset part or function, and the Researcher, who downloads the data and runs the FE calculations.

The prototype was demonstrated to stakeholders during the weekly meeting Cardio Vascular Group at the Amsterdam UMC, which includes technology and clinical researchers with a strong interest in acute care. We simulated the usage of the interfaces, for example, as shown in Figures 5-6. 

Figure 5: Interface for the researchers and principal investigators to download the research dataset

 

Figure 6: Interface for the researchers and principal investigators to perform calculations using the FE service

As the concepts used in this use case are rather new, only after seeing the demonstrator in action the researchers could fully understand the potential. The feedback received indicated that actually, the researchers want to use this type of service “now”, as this would reduce the current delays experienced for the delivery of the dataset for research. Also, the researchers were very excited about how easy it would be to select only the authorised features and download them in a structured way. Finally, the researchers became very interested in the FE concept and raised many questions about its extension to more complex calculations.

1.1.3      Where to from now?

Cybersecurity technology is of complex nature, so it takes quite some awareness, education, and even convincing effort to make new solutions come through. The demonstrator builtin ASCLEPIOS facilitated the communication about new solutions immensely because people could see and use them in a concrete system. For example, the ambulance services we interviewed understood the vision and became very interested in the approach, so much that they published an article about the project in their Vakblad V&VN AMBULANCEZORG edition 43, March 2022. Biomedical researchers, who are interested in data reuse, also became enthusiastic about the new encryption approaches that might help them unlock more data for research in a secure and privacy-preserving manner. There is a long path ahead before the vision comes to the users’ fingertips, but we plan to exploit the acquired knowledge and the demonstrator built for ASCLEPIOS for further dissemination and new research – stay tuned!