A few years ago, a technologist in a healthcare organization’s IT group would view data in the cloud skeptically, if at all. Today, that position is fading fast, with data to support it: IDC estimates that the healthcare industry’s spending on cloud technologies would surge to $13.6 billion in 2019.
Let’s explore what’s behind the surge, and the benefit to organizations that move to cloud-based data management. To illustrate, let’s outline a case of enterprise healthcare data management.
- In the first phase, a data lake, or a big data repository holding raw data in its native format, will be created. If data isn’t managed properly, it can quickly turn into a data swamp, providing meager benefits. Making it harder is the fact that healthcare data has layers of complexity that don’t lend themselves to the informal governing process that a data lake typically needs. To make the data actionable, and stay out of the swamp, each link in the healthcare data value chain must be managed, including: Data mastering – Healthcare organizations have many data sources that require mastering of patient and physician records to ensure that a golden record gives a single and accurate accounting of the patient or physician journey. Nothing will stall a conversation about quality faster than showing a physician patient data for whom they didn’t provide care.
- Data harmonization – In addition to many disparate data sources inside and outside the hospital, healthcare has hundreds of code sets, representing millions of codes where harmonization provides accurate information across data sources. As the National Institutes of Health (NIH) explains, some of the terminologies that often require mapping contain information not usually required for coding. Presenting a coder with this mapping task may make them feel a bit overwhelmed with the presentation of terms in an unfamiliar format. For example, a mapper who has the task of mapping a LOINC term to SNOMED CT may see something like this:
This is a common scenario, requiring the mapper to have a detailed understanding of the structure and content of the mapped terminologies. Research and study is needed in order to differentiate between the two structures and understand how they relate. Therefore, automated data harmonization leveraging industry and organizational code sets is an important step for creating usable information.
- Value set mapping and management – Value sets, also referred to as code groups, help healthcare providers and payers define clinical concepts, such as their diabetic population of patients. Each value set is a “bag” of codes which represent a particular disease or a type of medicine. A value set consists of taking terms, and their associated numerical codes, from standard terminologies such as ICD-10, SNOMED CT, RxNorm and LOINC. Value sets have a number of use cases, including creating clinical quality measures for CMS, defining a patient population cohort, defining decision support rules and developing application pick lists. For example, a health system could group together all of the ICD-9, ICD-10, SNOMED-CT and RxNorm codes that could indicate patients with diabetes. In this case, “patients with diabetes” is the clinical concept defined by the code group.
These are not the only steps in the healthcare data value chain, but their complexity shows where the gaps are in the traditional approach of moving data from a common place without structure. That traditional method pushes the work to the business intelligence (BI) analyst who must now navigate a data swamp, where the information is still in somewhat of a raw format and not necessarily optimized for analysis. This person also needs to take on the healthcare data value tasks required to deliver actionable insights which drive higher-quality care and reduced costs.
To reduce complexity and many hours of manual labor, a platform to curate a single version of truth is required. Gartner analyst Laura Craft has identified the emerging market need for this type of platform which she has coined, “Health Data Curation and Enrichment Hub.” “Formerly called ‘The Health Data Convergence Hub,’ this is the technology capability that brings together data from across the consumer/citizen/patient health and wellness continuum, and prepares the data for delivery to downstream consumption platforms, applications, analytics and “things.” It automates the ingestion of data – both structured and unstructured – from all identified and permissioned sources; provides tracking and traceability; and manages identity, compliance, and security. It may process algorithms and deliver the output to the correct modality.”
For healthcare organizations, this curated data layer could reside on premises or in the cloud. Let’s look at some of the advantages of managing the single version of truth in the cloud:
- Scalability/Flexibility – Long-touted as an advantage in the cloud, this is the ability to dynamically scale computing resources as necessary to meet and organization’s needs. There are a few practical implications of this advantage, too. When a healthcare organization first creates an enterprise data layer, it needs to ingest and manage many years of data, which sometimes creates a resource bottleneck. With cloud computing power, those resources an be scaled up, and then back down, once the project is back to updating daily data flows. Likewise, other data inflection points, such as a merger/acquisition or new data source (i.e., claims from a partner) may require a temporary increase in capacity.
With the move to 5G on the horizon, so is the rise of medical and healthcare IoT devices, which will dramatically increase the amount of data an organization handles. As the amount of data and data types grow, the associated complexity and the number of possible correlations for analysis also grows. You can then provide meaningful insights by managing and correlating this data back to the patient journey.
- Artificial Intelligence (AI) and Machine Learning (ML) – Large cloud providers such as Amazon Web Services (AWS) offer AI and ML capabilities which enable healthcare companies to learn more about patients and populations. Some types of ML – specifically unsupervised learning and deep learning – can identify patterns previously unseen by humans. AI/ML data that is cleansed and accurate will provide the best results.
- Security and Compliance – Premium cloud providers, including AWS and Microsoft Azure invest more in third-party security than any individual company using those solutions is capable of doing for itself. Basic services offer some security, though healthcare organizations in general need additional security, and HIPAA compliance, to meet their needs.
- Reliability – Cloud environments are redundant; if one location fails, another can take its place. This minimizes data loss and compute disruption (which can be a major time and cost factor when solving a complex, data-intensive problem).
- Real-Time Data – As the healthcare industry has become more digital, its ability to process information has accelerated. Rather than tracking patient data in batch format, organizations are increasingly monitoring patients in real-time to affect faster intervention. This improves patient health outcomes and can help lower the overall cost of patient care.
- Cost – Cloud computing costs are more cost effective than adding additional on-premise infrastructure.
- Latest Technology – Providers invest in the latest technology to stay competitive, while companies relying on their own resources must amortize the cost of existing resources; only investing in new technology as strategically and financially prudent for their own use. Additionally, in a managed cloud environment, the data architecture infrastructure is easily upgradable to newer versions.
- Architectural Flexibility – Healthcare organizations can add cloud computing resources to improve the ROI of workloads across its architecture. Beyond the additional data streams previously discussed, flexibility for increased capacity during peak reporting usage at certain times of the month requires minimal effort.
As healthcare approaches an increasingly digital future, enterprise data management will become imperative to success with new payment and care delivery models. Creating and managing an organization’s master version of their “true” data in a cloud environment in the present will pay dividends in the years to come.
 International Data Corporation (IDC). “Worldwide Semiannual Public Cloud Services Spending Guide”, August 2018
 Gartner, Hype Cycle for Healthcare Providers, 2018, Laura Craft, Mike Jones, July 2018