With their scale and flexibility, cloud computing technologies have the potential to open up long-sought-after opportunities for data collaboration in the healthcare industry. Traditionally, data platform architectures designed for pre-cloud computing environments have constrained the ability of healthcare organizations to collaborate. Some of these platforms have been repurposed and made available as cloud services, but their underlying architecture continues to limit access to all of the data available to drive better patient outcomes.
A cloud-first data platform can help healthcare organizations overcome obstacles in the collection, analysis, and sharing of data, and drive meaningful breakthroughs in evidence-based collaboration. Here’s how.
A large majority of American adults support increased access to health information for patients and providers, according to a recent survey for The Pew Charitable Trusts. A whopping 81% said different providers should be able to share health data about patients they have in common. But this is often easier said than done. Patient data originates from a myriad of sources and is delivered in a wide range of formats. Adding another layer of complexity, industry efforts to standardize around a common data model have emerged for certain subsets of patient data. As a result, it has become nearly impossible to collect all patient data into a single source of evidence, impacting downstream collaborations both within and between organizations required to confidently isolate the variables driving patient outcomes.
- The Observed Medical Outcomes Partnership (OMOP) format offers a standard for reimbursement claims and electronic medical records (EMRs).
- Regulatory agencies such as the U.S. Food and Drug Administration (FDA) require clinical study data to follow the Study Data Tabulation Model (SDTM).
- Unstructured data from diagnostic equipment follows various standards, such as the Digital Imaging and Communication in Medicine (DICOM) format for CT and MRI scans.
- From genome sequencers to activity trackers, the Internet of Medical Things (IoMT) generates data in a number of semi-structured formats that do not “fit” into traditional relational databases, such as JSON, XML, and VCF.
Cloud-based data platforms are solving the data collection problem by leveraging the flexibility and scalability of cloud storage to consolidate all structured, semi-structured, and unstructured patient data into a single, central repository and making it all available through a single, simple interface. Cloud object stores will hold everything from highly structured CSV to semi-structured VCF data to unstructured DICOM files in one place at a low cost, on-demand, and at near-infinite scale. The challenge then becomes accessing diverse patient information to generate the analytical insights required to go from data to evidence.
As the sources and structure of patient data diversify, it is not enough to simply unify disparate information into one repository. Conclusive insights in which a multitude of factors drive outcomes are dependent on a comprehensive analysis of all available data. The challenge with traditional relational databases is that they often require the transformation of rich, relevant patient data prior to analysis, which introduces complexity and barriers to access. In addition, a tight coupling of storage and processing power—which is typical of many databases architected for fixed-capacity, pre-cloud environments—often requires additional infrastructure to be provisioned to accommodate changes in demand from new or concurrent explorations of the data. Again, this introduces operational complexity, cost, and barriers to data access.
Data platforms designed to take full advantage of cloud storage and computing resources are addressing these technical limitations. From telehealth applications to connected medical device data, rich semi-structured patient information can now be instantly accessed and analyzed at scale using SQL—without the need for pre-processing and transformation steps—alongside traditional structured information such as prescriptions and claims data. A full logical decoupling of storage from compute resources allows these resources to be scaled independently and instantaneously, eliminating historical barriers and complexity to data access. This is particularly relevant for bleeding-edge techniques in machine learning and AI, which require relatively large amounts of processing power to execute and have historically required dedicated, expensive, and high-maintenance hardware.
At the heart of value-based collaboration is the secure, governed exchange of and joint access to patient information. But traditional data platforms pose significant limitations on the volume, frequency, and utility of data shared. Whether via APIs, file transfers, or a spreadsheet attached to an email, the mechanisms employed for data exchange largely rely on legacy processes involving physically copying and moving data. Data “hops” associated with copying and movement introduce latency, which erodes the value of high-frequency patient data from connected devices such as physical-activity trackers. Creating multiple copies of the same data in different locations introduces complexity with respect to versioning and maintaining a referenceable “golden record” as different parties manipulate separate copies. And from a patient privacy and data governance perspective, the prospect of needing to physically copy, move, de-identify, and protect sensitive personal information has been an obstacle to data sharing altogether.
Cloud-first data platform architectures are enabling a radically different approach to information exchange by eliminating the requirement for physical data copying and movement while maintaining access to the scale and flexibility of cloud computing resources. For example, the COVID-19 Research Database draws on data provided by over 30 U.S. healthcare and technology companies—from claims and electronic health records (EHRs) to lab and demographic data—to deliver a secure, anonymized patient data set made available for free to public health and policy researchers while allowing the healthcare and technology companies offering their data to maintain HIPAA compliance. A number of medical evidence information exchanges are emerging to blend together traditional assets such as prescription, diagnostic, patient history, claims, and reimbursement data with a rich range of alternative assets. Collaborators are invited to participate in the exchange, where they can then request and be granted secure, granular, read-only, or read/write access to specific assets.
For healthcare organizations to succeed at driving better patient outcomes through value-based healthcare, they must be able to glean insights and collaborate on data from a variety of sources without obstacles. Leading healthcare organizations are leveraging the modern data capabilities of Snowflake’s Healthcare and Life Sciences Data Cloud to drive value-based healthcare collaborations. To learn more, read our white paper, The Healthcare & Life Sciences Data Cloud for Real-World Evidence.