The Power of Analytics on Federated Data in the Healthcare Industry
The healthcare industry has always had its share of data challenges. While there’s a lot of potentially usable data available, it’s often fragmented across systems that can’t easily communicate with each other – or are kept “private” and never revisited. Security and privacy concerns over patient data are some of the major roadblocks to leveraging these disparate data sources for generating new clinical insights.
There have been a number of technologies that promise to overcome the challenges of clinical data sharing, but the role of artificial intelligence cannot be ignored. Analytics and machine learning on federated data is the first solution that can be realistically implemented on a larger scale. This method of computational data analysis is already being used by many hospitals to safely share their valuable data with research groups.
Read on to learn more about federated data in healthcare, and how SAIL is making clinical data available in a privacy-preserving and scalable way:
How Can We Securely Analyze Federated Data?
We’ll start with a more technical definition, and then explain in simpler terms.
Performing machine learning and analytics on federated data is a secure and privacy-preserving method of computational data analysis. It accelerates the speed at which AI models learn because the data is loaded into a single secure node for computation, a virtual machine that provides security to the data when computations are being performed. The secure computation node helps reduce the ability for a cloud provider operator and other actors in the tenant’s domain to access code and data while being executed.
In short, this methodology decouples the access of disparate data sources from actually performing machine learning on them. When provisioned by researchers, the secure computation node fetches encrypted data sets — meaning those researchers only can access the data they need for their specific request. Additionally, each participating hospital can’t access or see another participating hospital’s data.
So what does this all mean for a patient advocacy group?
Each hospital can keep its datasets cryptographically contained, so it is fully controlled and auditable at all times. It can’t be copied and pasted, emailed, forwarded, downloaded, etc. to a point of untethered proliferation, ending up on unauthorized servers.
Mistakes happen, even without bad actors. Perhaps someone downloads data on their laptop while using public Wi-Fi at a coffee shop. There is only so much a contract or an agreement can actually prevent.
Performing machine learning and analytics on federated data ensures that scenarios like this aren’t damaging.
Federated Data and Healthcare Research
This AI-fueled approach to federated data has the potential to make clinical data more representative, accessible and secure. Patient advocacy groups are excited about this new future.
There have been a lot of innovative attempts to solve the data challenges in healthcare, but so far most of them haven’t been entirely successful. Some promising solutions were electronic healthcare record (EHR) systems, health level 7 (HL7) standards, programs like SEER data, and even blockchain.
Performing machine learning and analytics on federated data is something that can be widely adopted throughout our healthcare system. IT teams will appreciate that data is only accessible while computations are being performed. Data from different hospitals can be cryptographically contained until it is needed for computation – at which point it is loaded into a secure computation node decrypted, and computation is performed. The computation node and hence the unencrypted data exist for the duration of the computation, after which it is shut down and data remains in storage in an encrypted state.
In addition, this approach can enable a new level of data access that fuels clinical research. Instead of requesting every hospital to share its data individually, researchers can quickly run machine learning algorithms against the federated data for multiple hospitals at once. This opens the door for one publication to be used for subsequent studies, leading to ground-breaking research that wouldn’t be possible without collaboration between multiple hospitals and researchers.
The federated data and AI approach goes hand–in–hand with representative and personalized research. Researchers will more easily be able to access datasets that include information on underrepresented populations or rare conditions. More exposure to this type of data means clinical research can be more inclusive.
Federated Healthcare Data with SAIL
Secure AI Labs (SAIL) is a next-generation clinical data registry that helps patient advocacy groups advance medical collaboration. By connecting hospital, patient advocacy groups, and clinical researchers, SAIL can enable healthcare practitioners to safely and securely collaborate and share research data. More importantly, the SAIL platform keeps patient data privacy at the core with a secure trusted execution environment, access controls, encryption and other privacy enforcing technologies.
SAIL’s Data Federation Platform utilizes proprietary AI technology that leverages machine learning on federated data that allows researchers to run machine learning models on hospital data normally restricted by privacy laws. This provides clinical researchers with new and diverse data sets to fuel their research efforts. In fact, researchers can more efficiently train their machine- earning models with previously siloed clinical data to make breakthrough medical discoveries.
Hospitals store their patient data in a secure enclave that’s inaccessible by outside sources. The SAIL platform can run machine learning algorithms against this data and return the results — and the secure computational node reduces the ability for any unauthorized people to access code and data while computations are performed. This is a highly secure approach to data sharing that ensures patient privacy while scaling innovation in clinical research.
Our registry helps patient advocacy groups gain access to an ever-growing volume of real-world clinical research data. We’re constantly onboarding new data partners with current and diverse patient data that you can leverage to expand your research mission, open up researchers to sources of more representative data and accelerate your efforts with machine learning. Join SAIL in overcoming the barriers to data access in the healthcare industry to improve patient outcomes for the future.
Schedule a demo to see how SAIL is bringing PAGs and researchers together to find disease-ending insights.