Last year, I wrote this blog post about the great potential of big data to drive innovation in healthcare. With the rapid progress organizations are making with their population health strategies, I thought it would be a good time to revisit this topic because of its importance to the industry. To be honest, healthcare organizations are still in the early adoption phases of big data. Only a few organizations are dealing with petabytes of data, which is typically the threshold people think of when it comes to big data. However, I do see significant progress. Big data is in the process of transitioning from a research-oriented activity to a main stream agenda item that enables multiple population health scenarios. Let’s explore why.

Population health is driving the data explosion

Whether healthcare organizations are ready for big data or not, many are faced with exploding amounts of data at their disposal as they embark on new population health strategies. Within a single health system, it is not unusual to have several hundred IT systems and applications in the portfolio, and to have 50-75 that are actually exchanging data. Population health often requires a better understanding of your patient population, utilization, costs, quality, and chronic conditions across multiple systems and or with many partner organizations. This likely involves extending beyond your EMR to other disparate EMRs and IT systems outside your enterprise. Therefore, one needs to aggregate, normalize and share clinical, operational and financial data from many IT systems across the community—including EMRs, billing systems, payers, pharmacy systems, labs, and HIEs. Also consider that today, most of the data sets in the industry tend to be structured. We are now beginning to see early use of unstructured data (i.e. clinician notes), which can contain highly valuable patient information (e.g. ejection fraction for patients with congestive heart failure). Additionally, I see increasing interest in other new data sets such as consumer, genomic, demographic and social data (e.g. fitness devices, purchasing history, Twitter, Facebook) integrated into scenarios for population health. There is no end in sight for the growth of data in healthcare, which I find incredibly exciting because of the ability to draw new value from it.

The mechanics of big data are being established

In order to realize the full value of their data, organizations have been establishing the mechanics of how that data needs to be aggregated, transformed and stored. It’s important to have the right population health data platform in place that can automate numerous processes, or else big data efforts can struggle to get off the ground. Healthcare organizations leading the way in big data have been adopting data platforms for healthcare that have the following capabilities:

• Automated data ingestion from any originating source in any format in real-time that has existing pre-defined configurations or parsers for a variety of data formats, e.g. HL7, CCD and CCLF, and unstructured.
• Automated transformation of data regardless if data is in different terminology code sets or structures that includes semantic mapping for code conversion and pre-built tooling that facilitates normalization and deduping of data from pre-defined and custom sources.
• Automated modeling of data that uses pre-defined healthcare entities that can accommodate most clinical and claims data as well as custom entities to accommodate customer-specific data.
• Open data sharing APIs that provide untethered read and write access with source systems and the ability to share data with analytics and self-service reporting solutions.
• Role-based security with auditing.

Initial population health scenarios are being explored

The scenarios for big data are limitless and I believe that one day they will expand to untapped data sources such as social media, consumer purchasing, and even things like smart clothing. Until that day arrives, big data can still have a significant impact on how you manage a population at an aggregate as well an individual level. Here are a few initial scenarios that organizations are already exploring today:

• Build and share a true longitudinal patient record to see all relevant patient data (e.g. labs, pharmacy, claims, analytics) that covers the full continuum of care.
• Employ predictive models (i.e. risk stratification), which are especially effective with large data sets so that you can focus your resources where they will have the greatest impact.
• Expand the scope of your analytics to new areas such as measuring the quality and financial performance of individual clinicians as well as overall organizational financial and utilization analytics.

It’s fantastic to see that healthcare is innovating through big data and population health. The organizations that have built their infrastructure strategically should be in great position to keep refining what they’re doing today while scaling and building new use cases. If you’d like to discuss how Caradigm can help with your big data strategies, then leave us a note here.

It’s a fantastic sign for the healthcare industry that the Fast Healthcare Interoperability Resources (FHIR) standard is garnering a lot of recent attention. I’ve had conversations with several CIOs who are hearing that FHIR could be the next-generation standards framework that can help innovate data sharing at their organizations. I can understand why they’re excited. EMRs have been closed systems for a long time, which creates major challenges for organizations wanting to share data between disparate systems. The challenge is even greater for organizations engaging in population health initiatives because the open sharing of data between different systems and providers is a must. Provider organizations want to learn more about FHIR because it has the potential to help overcome these interoperability challenges.

Let’s explore FHIR a little deeper. There are a few important details to know in order to understand its potential to help:

FHIR is an evolving standard

Up until September 2015, there were two active versions - DSTU1 & DSTU2. DSTU2 is the new version, but some technology vendors are still using DSTU1. HL7, the creator of FHIR, is still working on a final standard expected to be released in 2017. Once the final standard is in place, it will likely take a few years for broader adoption.

Vendor approaches vary

Solution vendors are in the early stages of developing FHIR strategies. Some EMR vendors are prototyping the use of FHIR APIs to enable read access of certain resources from patient charts. Others are working on the ability to read and write data back into systems. The types of data models that can be accessed using FHIR APIs also vary. For example, one vendor may support Patient, Allergy and Medication data models, but may not support Family History, Immunization, CareTeam or PlanofCare data models. Vendors can also vary whether they enable just read or read and write for each of the data models.

The use case is discrete data sharing

FHIR was designed for discrete data sharing, i.e., sharing of small batches of patient data. If you need to share one or two or ten bits of data, then FHIR can help. It is not intended for high volume data ingestion required for large scale aggregation.

FHIR is not the only game in town

Web services with REST-based APIs can already accomplish what FHIR seeks to achieve. Our customers don’t have to wait for FHIR, they can solve their data sharing challenges today using our rich and open web services data connectors that include role-based security controls and auditability. Caradigm has built unique applications like Knowledge Hub that can share real-time data and information from third-party applications directly within clinician workflows in their EMR. In the UK population health market, we have built mobile applications using REST-based APIs that can share patient data pulled from multiple sources to a mobile application at the point-of-care.

Caradigm fundamentally believes in open standards, data sharing, and in democratizing information to drive innovation in healthcare. That’s why we support an extensive number of data models, and have always been on the cutting edge of supporting emerging models such as NLP, unstructured models and now FHIR. We are able to engage services for FHIR API integrations, and will continue to build access to entities in the Caradigm Intelligence Platform including deeper integrations with specific EMRs. Caradigm also collaborates with other industry leaders on emerging standards by participating annually in events such as the IHE North American Connectathon Week (see this post about last year’s event). We look forward to participating in the next Connectathon in January.

Ultimately, there are many ways to approach population health, and Caradigm partners closely with our customers to understand their goals and challenges in order to help them develop strategies. We’re excited at the prospect of FHIR being part of the solution for our customers. If you’d like to discuss FHIR more and how it fits into your population health strategies, then please reach out to us here.

Healthcare organizations have become more aware of the need to leverage all of their data in order to support new population health management initiatives. An Enterprise Data Warehouse (EDW) is one of the key solutions many healthcare CIOs are considering to help accomplish this goal. In recent conversations that I’ve been having with CIOs, I often hear them say that they need more than what a horizontal EDW provides. The “Aha!” moment comes when CIOs realize that they need an EDW specifically designed for healthcare that has the vertical functionality needed to drive a scalable healthcare data and analytics strategy.

The first step in building an EDW for healthcare starts with choosing an enterprise EDW foundation. Caradigm has developed a deep integration with Microsoft SQL, a Leader in the 2014 Gartner BI Magic Quadrant that provides a strong horizontal EDW foundation including SSIS, SSAS, SSRS and Power BI tools. Caradigm has added an array of vertical functionality to that foundation to deliver an EDW for healthcare. Let’s explore further what distinguishes it from horizontal solutions.

A single-source of aggregated data in near real-time

Aggregating different types of data (e.g. clinical, claims, financial) from potentially dozens of systems across a health network is a core requirement for population health that horizontal EDWs are not equipped to handle efficiently. An EDW for healthcare is different because it provides the following functionality that enables a single-source of data to be possible:

• A healthcare data model that can automate the process of combining different data sources and data structures to create a single, longitudinal patient record.
• Complex healthcare data aggregation parsers that automate the ingestion of data from all healthcare information technology systems and normalize disparate data to semantic healthcare terminology tailored to your enterprise.
• The ability to use Hadoop and NLP (Natural Language Processing) to leverage and derive insights from non-structured data.
• The ability to update and make the data available in near real-time as opposed traditional EDWs that require delayed monthly or quarterly batch processing.

Actionable and Extensible Data

An EDW for healthcare also must deliver the following functionality that enables the data to drive action:

• The ability to write data back into source systems to surface actionable information at the point-of-care. This is a key requirement that allows you take action from insights.
• Healthcare specific tool sets for non-technical clinical analysts that allow them to perform analysis and reporting with strong visualizations. You want to decrease the barriers for information access by bringing end users closer to data. The traditional route of requiring end users to work through IT for coding reports is slow and expensive.
• Data exploration tools should enable insight discovery i.e. exploring data to discover hidden insights versus the traditional route of asking the questions and building rigid data marts around them.
• Native support for predictive analytics like estimations of risk and predicted outcomes. Examples include cohort stratification, patient identification, risk modeling, readmissions management, and total cost of care.
• Integrated out of the box analytics applications like Quality Improvement, Risk Management, and Condition Management that can leverage the EDW to perform and share analytics.
• An open platform that can share data via web services APIs (Application Programming. Interfaces) or access via other 3^rd party popular BI tools like Tableau, QlikView, and Spotfire
• An application development platform that gives the ability to create new applications utilizing the EDW.

Security and Compliance

Lastly, but as important as any of the functionality mentioned above is the ability to provide a security model that is role-based, row based, field level redaction with auditability. This can be an important tool for HIPAA best practices.

My advice to providers is that they need to think about which tools can help them today and scale with their future needs. The overall strategy has to be extensible and simple from the customer’s point of view. Providers shouldn’t have to acquire multiple new systems, develop custom solutions, or build an internal team of developers. Providers need a partner with a defined path forward that includes infrastructure, domain expertise, out-of-the-box functionality and tool sets that can simplify processes today while being able to adapt in the future. If you’re struggling to get out of the gate beginning with data aggregation, then that’s an indicator that there are missing fundamental capabilities. It’s unlikely that population health data capabilities can be patchworked together without delaying the timeframe for success and increasing costs.

Caradigm is unique because we deliver a mature and comprehensive EDW designed specifically for healthcare. We have already helped customers aggregate their data and are surfacing that information in clinician workflows to improve care. Once these core requirements are in place, providers are positioned well to succeed with their population health initiatives. I look forward to having more discussions with providers about how we can partner to help you realize the full potential of your data to support your population health efforts.