Medical digital twins secure COVID-19 data

Dell has partnered with the i2b2 tranSMART foundation to create privacy-preserving digital twins to treat the long-haul symptoms of COVID-19 patients. The project hopes to improve treatment for the 5% of COVID-19 patients who develop chronic health issues. The new tools integrate de-identified data — which refers to data from which all personally identifiable information has been removed — AI, and sophisticated models that allow researchers to perform millions of treatment simulations based on genetic background and medical history.

This initiative is part of Dell’s long-term goal to bring digital transformation across the healthcare industry. Jeremy Ford, Dell vice-president of strategic giving and social innovation, told VentureBeat, “AI-driven research and digital twins will support hospitals and research centers globally and contribute to Dell’s goal to use technology and scale to advance health, education and economic opportunity for 1 billion people by 2030.”

The i2b2 TranSMART foundation (Informatics for Integrating Biology at the Bedside) is an open-source open-date community for enabling collaboration for precision medicine. The group is focused on projects to facilitate sharing and analysis of sensitive medical data in a way that benefits patients and protects privacy. The partnership between Dell and i2b2 promises to create best practices for applying privacy-enhanced computation (PEC) to medical data.

I2b2 chief architect Dr. Shawn Murphy told VentureBeat that medical digital twins are essential because they enable “patients like me” comparisons across very large cohorts of similar medical twins. This will help identify things like biological markers for diseases and compare treatment options for patients who share similar features like age, gender, underlying conditions, and ethnicity.

Multiple sources and types of data go into constructing the medical twins, including a patient’s Electronic Health Record (EHR), consultation information directly from the patient, and waveform data from cardiac monitors, ventilators, and personal fitness tracking devices.

“They could be used in the future to help researchers perform millions of individualized treatment simulations to identify the best possible therapy option for each patient, based on genetic background, medical history, and a greater overall knowledge of the long-term treatment effects,” Murphy said.

Privacy required for adoption of medical digital twins

Privacy is a crucial requirement for the widespread adoption of medical digital twins, which require combining sensitive medical data to create the best models. “There is a significant amount of work to collect, harmonize, store and analyze the different forms of data coming from multiple locations while maintaining patient privacy and data integrity,” Murphy said.

Dell is focused on providing data management hardware, software, and integration services for the project. The data enclave was designed to provide the computational, artificial intelligence, machine learning, and advanced storage capabilities needed for this work. It consists of Dell EMC PowerEdge, PowerStore and PowerScale storage systems, and VMware Workspace ONE.

Researchers are still in the early days of identifying vulnerabilities in these architectures and balancing these against performance and workflow bottlenecks. With secure enclaves, sensitive data from various sources is encrypted in transit to a secured server, decrypted, and processed together. It assures the best performance and streamlines workflows of all PEC technologies, but also requires extensive security analysis because the data is processed in the clear. Other PEC approaches, such as homomorphic computing, can process encrypted data but also run much slower and are more challenging to integrate.

Murphy said additional infrastructure would be required to support new locations and expand the data pool to include research centers in minority institutions and hospitals outside the U.S. “This is particularly critical for the full representation of diversity in digital twins,” he said.

Building a common language

The digital twin research started with the creation of the 4CE Consortium, an international coalition of more than 200 hospitals and research centers including data collaboratives across the U.S., France, Germany, Italy, Singapore, Spain, Brazil, India, and the United Kingdom. The 4CE consortium brings together all the sources and types of data to create a ‘common language’ to enable comparisons between different sample populations. This allows comparing medical digital twins that share similar biological markers to see what therapies work most effectively for other patients in the real world.

In theory, researchers should be able to pull in data from the EHR, which is designed to manage all the medical history, including treatment options, medical appointments, diagnostic tests, and resulting treatments and prescriptions. However, in practice, Murphy said EHRs are prone to inaccuracies and missing information. For example, in the U.S., the code for rheumatoid arthritis is used in error four out of ten times when the code for osteoarthritis should be used. “This is why we need to aggregate multiple sources and types of data which in unison will spell out the condition of the patient,” Murphy explained.

The real value of the EHR comes when combined with real-world patient interviews and other forms of data to create medical digital twins and drive population-level insights. The technology used to understand long-term COVID-19 symptoms can also help create high-resolution, disease-specific medical digital twins that can be used by physicians and researchers for many other applications in the healthcare system.