December 27th, 2017 by Christian Seebode

In part one we learned about the requirements and the need to build a Patient Centered Infrastructure. In part two we learned about the Patient Centered Process. This article explains the need for profile information in the Patient Centered Process.

Profile Information as explained in the previous article references all the necessary actions to collect health data and access them again at any stage of a healthcare process or any step inside the Patient Centered Process. This already says a lot. And it contains a clear statement why a Patient Centered Profile is something very different from a traditional EHR Profile. EPR or EHR implementations are traditionally trying to collect healthcare data in a way that they act as data sinks for healthcare processes. Standardization efforts have tried to assure interoperability of these data oriented tools.(EHR Profiles)

The success was very limited. One fundamental reason for this is that healthcare itself is not a data based culture. Healthcare traditionally was and hopefully always will be a value oriented culture with its undisputable objectives providing and maintaining health of people and populations

Data collection to support these goals was done ever since to document experience. With a modern and more scientific medicine data became more important and the primary tool of proof for effects and success. However there is always this always this fundamental bias involved in any attempt to collects data from one patient, in one situation according to the question that needs to be answered. The dimensions of who, why, what, when, where and how are responsible for the establishment of any data silo, that is so difficult to break down again.

For data to represent valuable information it must reference all the knowledge of all these dimensions together with their concrete values along these dimensions. It is very often not understood how a data value is biased by the method of generating it or the time or the question behind it. Data quality almost always does not transcend one or more dimensions. This is a fundamental problem for any value based healthcare system that relies on data collection.

I don’t want to describe the problem any deeper than that. In fact is makes more sense to thinks about possible solutions. Well, the designated solution of this series is called the Patient Centered Infrastructure. Together with the Patient Centered Process it takes care of (re)assigning value to profile information. How this can be done will be topic of upcoming article. Here we just focus on the necessary aspects of profile information.

The Patient Centered Profile makes sure that all values are conserved together with the knowledge behind its acquisition. In fact that is not very simple. The expanded (and probably exaggerated and unnecessary) version of this claim is that the necessary knowledge to document a lab value contains also the science that led to the discovery of the method of acquisition itself and the concrete instance and method of the measurement. But more importantly the knowledge that leads to the understanding of the lab value needs to be referenced. That includes also the patient’s knowledge according to the individual level of health literacy. This is a powerful statement. In fact the central claim is in order to break down siloes it is necessary to (re)assign knowledge to profile information in a very agile and dynamic way. How this is done will be documented in upcoming articles. We will now have a look at how profile information itself should be managed.

The Patient Centered Profile contains data collections accumulated in the Patient Centered Process and supports all the necessary dimensions to interpret the data in a meaningful way rendering information to understand the pathway of the patient. Dimensions are supported as views or filters on the data. Semantic filtering can be used to document particular diagnostics or questions. The most necessary dimensions are who, why, what, when, where and how data was acquired and of course the knowledge necessary for understanding, but of course now restricted to these. Dimensions can be added dynamically. It is very important to attach the necessary knowledge explicitly. There is no such thing as implicit knowledge which usually is the main reasons to build unbreakable data siloes. Doctors and patients should be able to look at the Patient Centered Profile using different levels of knowledge but still referencing the same data in a way that a Patient Centered Dialogue is supported. An example for using different relevant knowledge or at least emphasizing different views would be a red blood cell count which for the doctor is related to the condition of anemia and for the patient it is related more to the clinical symptoms like fatigue and of course the treatment depending on the type of anemia. Even if both use different working knowledge concepts attached to the same data, they are enabled to cooperate on that condition.

This means that a semantic patient profile or patient record always has the working knowledge attached to it that helps to interpret the data according to the health situation or condition. This is very important for rare conditions where you have to collect common conditions which usually lead to common interpretations in a different way usually hidden form the view. An example for this would be Fabry’s disease where many symptoms are caused by a rare genetic defect and not by very common diseases. In fact a semantic profile supporting different views on the data is the key to effective translational medicine.

This article only motivates the fundamental data structure for supporting semantic profiles within the Patient Centered Process. Upcoming posts will document the necessary processes and intelligence that manipulates semantic profiles in a meaningful way. A Patient Centered Profile supports different views on the data as already mentioned. A Patient Centered Profile is constantly updated as new data or knowledge flows in. These different versions along the time dimension can be compared and manipulated in a transactional way to support differential analysis.

Conclusion

The patient Centered Profile collects medical data from unlimited sources and stores them together with the knowledge concepts attached that are needed for interpretation according to a certain question or condition. It supports various views and different outcomes along multiple dimensions and can be updated and manipulated to support different views on the same situations. The knowledge management and mining processes will be documented in upcoming posts

Comments are closed.