Intelligent Healthcare part 4 – Process Management & Modeling

In our last post on Intelligent Healthcare, we talked a bit about Electronic Healthcare Record systems. EHR/EMR technology is an important piece of the larger set of clinical systems as it represents a patient centric organizational framework. However; EHRs are only part of a larger picture. One area that is particularly promising for the application of Semantic technology to healthcare is process management. When we discuss process management in this context, we’re not talking about traditional process management software solutions. Healthcare process management is in a sense a formalization of (medical) practice approaches that for the most part aren’t automated and in many cases likely never can be fully automated.

Healthcare, by its nature, requires a good deal of physical interaction – all of that interaction though is guided by various formally or informally managed processes including:

•    Diagnostics
•    Prognostics
•    Treatment
•    Patient Tracking
•    Facilities Management

The body of knowledge that practitioners have as potential reference is quite daunting; even with 12 years of education plus continuing training it is impossible for any one practitioner to know or remember all of it at any given time. Care providers must rely on a combination of well-defined procedures as well as the ability to recall some of that larger storehouse of information when needed. Decisions often must be made very quickly; there may be no time for extended examination or reference reviews, especially when there are many patients and relatively few caregivers. So, the challenge is how do we make the procedures easier to reference or manage and allow for more efficient leveraging of the larger body of medical knowledge in everyday practice without interfering with the efficacy of current practices.

This challenge implies a specific “Problem Set” (a meta-category of related aspects that together comprise all elements of a problem) associated with the technical resolution (or realization of Intelligent Healthcare).

The initial problem set for Intelligent Healthcare includes:

•    Ability to capture and represent processes (procedures).
•    Document-centric, message focused integration.
•    Conflicting (Healthcare-related) terminology.
•    Conflicting standards and resulting implementations.
•    Exclusion of free text data in traditional EHR/EMR systems.
•    Lack of specific support for Healthcare practices / processes.
•    Security & Data Integrity of exchange transactions.
Intelligent Healthcare Challenges
The diagram above illustrates that the nature of challenge for Intelligent Healthcare is complexity, both in terms of medical practice and the ability to achieve interoperability between clinical systems. Both EHR technology and semantic process automation begin to solve this challenge. The key takeaway from the diagram is the realization that automation without interoperability makes healthcare more complex than before because we are necessarily adding many new sources of data to nearly every medical process without the ability to correlate it in meaningful ways. This adds a burden to the caregiver when in fact the systems should be facilitating the existing knowledge overload crisis.  

So what is a Semantic Process or representation of a procedure? In current medical practice many caregivers develop cognitive tools referred to as diagnostic algorithms. This type of tool is in fact in most cases a decision tree. The diagram below gives an example of what one looks like, although there is no standard format. This diagram is a diagnostic aid, but as you can see, the format could be easily modified to support prognostic decisions and treatment monitoring or outcomes. These types of diagrams can be represented using RDF triples in combination with one another. The data store that drive them can become a remarkable new resource for process-driven information and the way to allow this to help caregivers is to give them the ability to rapidly access this visual knowledge and create their own variations built atop evidenced-based care scenarios.
Diagnostic Algorithm
Where this gets very interesting of course is being able to place the processes or procedures in context to a unique individual, the patient. This requires interoperability between process and patient data (EHRs) so that caregivers can be given access to procedure options based upon test results or patient sensor monitoring. While many might expect that this type of capability may be provided in some sort of integrated toolset, it is in fact better approached as an interoperability exercise. The knowledge of healthcare providers can become a shared, web-based resource; one that can be used anywhere and is grown by the practitioners themselves. We will discuss this in more detail in our next Intelligent Healthcare post.