Better Information Management, Better Healthcare Delivery

One of my recent posts ("Cloud, Big Data and Healthcare") was a bit more popular than I expected.  

More than a few people contacted me to share that one of the most important (and interesting) areas in healthcare delivery was around the building and exploitation of VNAs -- vendor neutral archives. 

Their case was simple: through the widespread adoption of VNAs, we'll have ultimately better metadata to analyze, and hence an expedited pathway to big data analytics making a meaningful impact on how healthcare is delivered. 

In some sense, it's about finding logical "gravity wells" where useful metadata can be gathered, coalesced and made productive.  

And large image repositories in clinical settings seems to be a leading candidate. 

Welcome To The World Of Medical Imaging -- And Medical Information

One of the most important (and challenging) information types in clinical environments is medical images.  This particular data type seems to be exploding along each and every axis: more image types, more reliance on them in diagnostics, more frequently used, far greater resolutions, moving images vs. static images, requirements to keep them around longer, make them immediately accessible when and where needed -- the list goes on and on and on. 

For someone who's accustomed to tossing around large storage numbers, the capacities and forecasts coming out of this particular segment are impressive indeed :) Originally, these information repositories were the sole province of PACS vendors, usually as an adjunct to the equipment that was capturing the image, such as a CAT scanner. 

Thanks to diligent and persistent work by the DICOM effort (and the related HL7), there now appears to be a usable and widely accepted standard for separating the application from the archive.  

Doing so enables the imaging information to be used in much more productive ways other than simple, isolated diagnostic tests. A key part of the evolving DICOM standard is its support for extremely rich metadata -- I've heard there are over 1500 fields defined, and probably many more in the pipeline.   

Unlike so many industry standards, the DICOM group appears to be getting the job done for the people who use the technology vs. the vendors who create it. 

Here's the hope: since medical imagery is one of the more important (and costly) information types underpinning healthcare delivery, it will undoubtedly attract attention and investment by supporting IT organizations.  

In the process of doing so, the metadata contained, managed and improved through DICOM-compatible applications yields a likely foundation for the next wave of big data analytics.  Fix the problem for medical images; you've got a foundation that can wrap everything else around it.

Forcing Functions? 

A recent Gartner note gives support towards rising interest in VNAs (vendor neutral archives) by clarifying what qualifies, and what shouldn't.  The defining characteristic, not surprisingly, is neutrality in light of traditional PACS vendors. 

Predictably, Gartner leans towards the rationale for DICOM-based VNAs as being cost motivated.  While that is certainly a strong incentive, I would believe that the more powerful motivation is "doing more with your information" -- at least, more than your PACS vendor might have had in mind. 

It's also helpful to point out that healthcare delivery (in the US, anyway) appears to be federating and consolidating.  Healthcare providers are joining forces to offer better services with more efficiency.  In this process, it's not uncommon to find two or more PACS-like applications that don't do a good job of sharing with each other.   Again, another motivation for the VNA concept. 

The EMC Approach 

Historically, we've played in this market by partnering with the PACS vendors around storage requirements, e.g. Centera and the like.  While that continues to be a reasonable way to participate, we believe we can do much more to help healthcare delivery organizations build a foundation for their future. 

At the foundation, anytime you're talking massive metadata, strict policy, potential geographical dispersion and large objects, there's a case to be made for purpose-built object storage.  In the world of cloud storage and service providers, this is Atmos.  

While other styles of storage can play at the periphery, Atmos is central here -- almost designed for the task, if you look closely. As one example, important data needs to be protected and archived.  Or tightly controlled as to where it can - or cannot -- go. Or have audit trails of where it's been.

Many of the native capabilities in Atmos do this as a natural extension of the policy-driven services it offers, but -- as needed -- these can be augmented with our competencies in backup, recovery and archiving. 

Manipulating and managing metadata around content is something that Documentum does quite well in this context -- especially if there is non-image content being managed using this model. 

Around this image and metadata there's the need to support integrated workflow.  We've partnered with companies like Acuo Technologies to create a "DICOM grid" layer that can manage the flow of information across different applications and systems.  

Presentation and collaboration around the images themselves is also essential; one of our leading partners in this space is lifeIMAGE who offers a cloud-based service that's very popular. 

Finally, at the top of the stack is the need to support researchers and administrative professionals with deep analytics: hence the inclusion of Greenplum.  The more data sources available to this layer, the more value it will provide to the organizations that use it.  And DICOM-based VNAs look to be an important source indeed in the near future. 

We think there will be two primary delivery models for this stack.  For larger organizations with considerable IT investment, a private cloud approach looks to be a likely candidate for the supporting infrastructure and operational processes.  

Smaller organizations will likely be served by focused community clouds offering specific services tailored to their needs.  And a few larger healthcare delivery organizations are working towards providing these services on behalf of the smaller organizations in their region. 

Challenges?  Plenty! 

Hopefully you can see the potential, but what about the problems?   

First, there's the issue around patient confidentiality and associated information management regulations.  Healthcare delivery organizations are held accountable for the information entrusted to them.  But -- almost perversely -- the value of the information increases the more it's shared by healthcare professionals, and often across organizational boundaries.  

And, especially here in the US, we tend to value our privacy :)

Second, there are stakeholders with agendas here.  To control important information (or access to it) confers power, influence and leverage.  And, despite noble intentions all around, there's a pragmatic element to consider that people (or organizations) are often reluctant to "give away" something of value.  Even if it's in the patients' best interests. 

Aggregating and using metadata to support advanced analytics might sound like a huge challenge, but – in essence – that’s what data scientists and healthcare researchers do for a living.  I’m sure there are a few technology pieces that can help along the way, though. 

Then there's the maturity angle; not so much around the technology, but the operational processes that deliver it.  We sometimes joke about low-priority IT problems not being a life-and-death situation, but -- when it comes to healthcare delivery -- very often it *is* life and death we're talking about. 

And, finally, we're talking about a substantial journey here for larger healthcare delivery organizations.  Transitioning from the legacy information management model to the new model won't be quick, easy or cheap.  And, in a world of competing priorities and short-term fire drills, it's going to be hard for many to get the focus to move in this direction. 

Thanks to Ken Waldbillig of EMC for the quick tutorial here :)

The Broader View? 

Many of us in the technology industry have the privilege to look across all sorts of interesting industries and use cases for technologies.  The more you look, the more you see common patterns start to emerge.   

Separating application from information.  Investments in harvesting the information's value, apart from the application that created it.  Delivering useful and actionable information to knowledge professionals in a convenient, easy-consume manner.  Cloud-like delivery models with automation throughout.  Security and compliance built in, vs. bolted on. 

Where it gets interesting for me is how exactly these broader themes play out in specific industries and use cases.  The devil is always in the details. 

When it comes to healthcare delivery, especially in the US, it looks like the stage is set for a shift in thinking on how critical healthcare information is managed -- and leveraged -- for the benefit of us all.

 

Comments

[Dana]

Like the blog, appreciate the share!