In 2019, medical researchers at the Jackson Laboratory in Bar Harbor, Maine, were searching for a way to accelerate the development of cancer drugs. They faced a unique challenge. In a field that publishes hundreds of new research papers daily, the information needed to save a patient’s life might only be available in one paper. No detail can be overlooked, yet no team has the time to read hundreds of scientific articles every day.
Working in collaboration with Microsoft, the Jackson Lab developed the Clinical Knowledgebase (CKB), an AI-powered database consisting of published literature and curated information about genomic mutations and cancer drugs. Today, the CKB makes it easy for oncologists to discover precise matches between a patient’s genes, tumor markers, and treatments.
A privacy-preserving technology collaboration that delivers medical research breakthroughs
With large and diverse data sets, researchers can develop and validate AI algorithms whose results are generalizable to broader populations and clinical scenarios. However, data sharing agreements between researchers from different organizations can be difficult to execute due to concerns related to data security, privacy, and use rights.
While working at the Center for Digital Health Innovation at the University of California San Francisco (UCSF) and in collaboration with Microsoft and other partners, the founders of BeeKeeperAI developed and validated a sightless computing platform to address this issue. The company has since spun out of UCSF and developed a commercial software as a service product called EscrowAI to facilitate secure collaboration between algorithm developers and data stewards, allowing algorithms to compute sightlessly on privacy-controlled data using the Azure confidential computing environment and resources.
Working within a Zero Trust framework, EscrowAI maintains the privacy of patient data throughout the algorithm compute process. Data stewards retain control of their data at all times. By facilitating sightless computing, EscrowAI promises to remove the need for costly and time-consuming data anonymization. This dramatically reduces the risk of data breaches when allowing third parties to compute protected health information (PHI) data and avoids costly sanctions and tarnished reputations while maintaining compliance with Health Insurance Portability and Accountability Act regulations.
Another privacy-preserving data approach involves verifiable data provenance and granular personally identifiable information/PHI consent management. Equideum Health, working with Microsoft, assists cross-enterprise data collaboration without data centralization, using Azure confidential computing and Microsoft Purview.
In this approach, data from each collaborating institution is normalized, cataloged, and enriched with new classes of metadata related to provenance, identity, and granular consent. The expanded metadata enables cross-enterprise data discovery, which is used to structure compliant collaborative analytics and machine learning workloads governed by fine-grained, enforceable, and dynamic consent.
The collaborative compute workloads are then orchestrated across participating enterprises without exposing, sharing, or exchanging their respective source data. This removes the need for de-identification and, therefore, preserves complete data context, quality, and longitudinality. Trusted and verifiable analytical insights are then available to the participating enterprises.
These types of secure privacy-preserving approaches to data sharing and collaboration are particularly relevant to researchers who receive grant funding from the National Institutes of Health (NIH). Beginning in 2023, all researchers who receive funding from the NIH must make all the scientific data used in their NIH-sponsored research openly available.
The NIH’s objective is to accelerate biomedical research through the open sharing of high-value data sets. Zero Trust and Confidential Computing Collaborative approaches to data sharing can help ensure that clinical and research data is shared in ways that preserve privacy and intellectual property.
The NIH Science and Technology Research Infrastructure for Discovery, Experimentation, and Sustainability Initiative, or the STRIDES Initiative, has also taken a keen interest in data sharing. Working with Microsoft, STRIDES extends special pricing on cloud technology—along with access to specialized training and subject matter experts—to medical researchers.
Healthcare AI applications improve patient care, outcomes, and efficiency
Many AI-driven solutions have a direct impact on patient health. Take the case of a project that doubled the capacity of scarce medical equipment in 2020. Volunteers from Duke University used AI for Health from Microsoft to design a life-saving ventilator splitter as an emergency measure. Drawing on the Voyager-EUS2 supercomputer on Azure, which is the world’s fastest public cloud supercomputer, the team logged 800,000 hours of computing time in just 36 hours to optimize the project and bring it to doctors and patients across the United States. With the splitter, doctors could temporarily place two patients on a single ventilator.
Another healthcare AI project is underway at Rush University Medical Center, where physicians are using ambient clinical intelligence (ACI) to convert patient-physician conversations into structured clinical progress notes that can be seamlessly integrated into the electronic health record (EHR). With ACI, a clinician’s time spent documenting in the EHR is reduced by 50 percent. As a result, 70 percent of clinicians feel less burned out, and 83 percent of patients report a better experience. First-time approval for prior authorization also increased by 40 percent.
Advancing the future of medical and healthcare research
It’ll be exciting to watch how academic medical centers around the world will use cloud technology to increase life-changing research. Using Microsoft solutions, they’ll drive innovation to advance medicine and improve patient care and experiences.
For more information on these opportunities and partnerships, visit Microsoft Cloud for Healthcare.