December 14th, 2019 by Christian Seebode

In the preceding posts of this series we learned about the Patient Centered Infrastructure and all its constituents that ensure a seamless patient journey which always supports a precise picture of the individual health situation, collects all information about possible outcomes and helps to perform the necessary tasks in order to reach the personal health goal whether it is a cure, a remission or just an improvement in daily life.

We learned that profile information should be collected with the knowledge that it represents attached to it and that it should be versioned as knowledge is constantly changing in order to keep track of the consequences of change.

The next piece of conceptual model of the Patient Centered Process is information retrieval. Information retrieval uses the information backed up in the Patient Centered Infrastructure whether it is collected in profiles or provided with medical service or by educational means. 

What is information in this context and what is retrieval? 

Informational retrieval is a discipline that combines methods and results of computer science, information theory and other sciences to provide the theoretical background for searching and finding content. You can read more of it here: https://en.wikipedia.org/wiki/Information_retrieval

Information retrieval for the Patient Centered Infrastructure means to give patients access to information and to provide the ability to assess the quality of this information according to the personal health situation, the individual level of health literacy and how this information can be used in other steps of the Patient Centered Process. This is not trivial at all.

In order to retrieve information, you first need an interface that supports retrieval actions. A very common understanding of retrievel supports search options on a repository like for example a database. First important thing is to understand how to issue the search and the search criteria. Is it a keyword or a combination of some? Is it some other information structure like an image or la results? Or is it a complete document or phrases written or spoken in natural languages? Or is it something completely different? Well, I would say it’s all of them. Impossible, you’d say. Well, is it?

At this point it is not necessary to make any real world concessions to what is possible or not. We are just collecting requirements. We didn’t make concessions to describe the profile step in the previous post, we won’t do it here and we certainly won’t do it in upcoming posts. We are just collecting problems and requirements. Solutions comes next, if ever …

When I started thinking about Patient Centered Concepts and blogging about it on http://patient-centered-it.com/ healthcare was moving along well established paths. It is still the case but today you feel that lots of money flows into a change towards patient centeredness, in fact the whole industry tries to understand what it means and the value it creates.  What I am trying to say is that given the current technological possibilities or even concrete products we don’t need to limit our view because future healthcare will need a couple of inventions that remain to be seen.

Roughly speaking information retrieval in the Patient Centered Infrastructure does everything in order to give patients access to information in a meaningful way.

This doesn’t exclusively mean that the result of this retrieval is something patients must read. It could be retrieval as part of a workflow step, information that is required to do a dialogue or just some information that configures a personalized device. In fact information becomes a first order artifact in medical procedures and becomes as important as other components in the process including therapeutic agents. 

But how does this work? What is needed conceptually and where does this lead us? 

Let’s just start simple, such that possibly everyone may follow. Using search engines like Google is an everyday task. Whatever you are looking for you will get some useful hints within seconds. The user interface is so simple and smart at the same time that virtually everyone who can input information will get some relevant results. In case of a search engine you type in something like keywords, phrases or an entire sentence which the search engine than maps somehow to the relevant results. This example is only simple because everyone is used to it. The technology behind it is far from simple. But let’s not bother about how this is implemented but rather how it is modeled. The only important fact here is that Input and output is related to the knowledge of the user. The input is usually much more related to the user’s knowledge than the output because the input usually is coming from the user. Lets ignore the cases when it doesn’t for a while. In case of the search engine the input usually uses a textual representation, mostly human language, and the output also is a collection of texts or links to texts that are related to the input. Without digging too deep into information science here the most relevant concepts of this retrieval process are the input or the query and the result in its most abstract sense. Contributing components are also the knowledge or literacy of the user and the model, in its most abstract sense, of the search engine. However the most important relationship remains, like the input is related to the knowledge or literacy of the user, the relevance of the output is it too. A relevant output should always have some meaning for the requester, if it has not it is just useless, even if it is correct. 

Now let’s move away from this simple, known example and bypass all other retrieval technologies and models like databases of all kinds and step directly into the Patient Centered Infrastructure as defined by this series of posts. 

What does it mean to query information from a Patient Centered Infrastructure? How is information represented, processed and how results are represented? Well, there’s not a single and simple answer to that. Stepping back to the posts about Profiles in a Patient Centered Infrastructure this means at least that we can use Profile as an important source of information. Additionally of course all data sources that are linked with the Profile may be queried too.

But we need to look at the use cases for information retrieval first before we can understand what is needed and more important how it is needed. Information retrieval from a Patient Centered Infrastructure will always be embedded in a higher level use case like engaging with a medical service, communicating with other stakeholders or making decisions about possible process steps. The use case defines not only the domain and the vocabulary but also the process and interactions. To put it in a simple way without the need for formal definitions, which are of course necessary for a sound scientific model, information retrieval from a Patient Centered Infrastructure means to access just the right amount of information as required by the current use case or interaction with the infrastructure. This could be pretty simple like just asking for a definition of a word or it or it could be amazingly difficult like calculating the recommendation of the next best step like adjusting a medication dosage or selecting the best choice for an interventional procedure in the given situation, or just the next best available person to talk to.

The Patient Centered Infrastructure itself has no predefined knowledge about how to retrieve information and less even what it means (see upcoming post about knowledge management), it only provides the infrastructure to connect use case provider and consumer together in a pretty dynamic fashion. 

Before we take a look on what is needed to realize this pretty big picture let’s have a look on the requirements. 

Patients should be viewed as active stakeholders in consumer health information technology development (http://www.jmir.org/2018/4/e128/) and this is true for all stakeholders of the Patient Centered Infrastructure. This requires a very dynamic, declarative or data-driven approach about how to define requirements, interactions or processes. Trust in health information and belief in the effectiveness of information security safeguards increases perceptions of patient care quality (http://www.jmir.org/2018/4/e107/). Safety and trust is a first order requirement for the Patient Centered Infrastructure in order to make it work. It has to be very transparent and concise about safety levels and even the comparison with perceived safety. WhatsApp, SMS, and email are effective and widely used ICTs that can promote communication between cancer patients and physicians (http://www.jmir.org/2018/2/e50/). The adoption and integration of everyday ICTs is a big factor for popularity and ease of use of the Patient Centered Infrastructure. Health care professionals should assist patients’ medical decision making by initiating as much dialogue with patients as possible, providing credible and convincing health information to patients, and guiding patients where to look for accurate, comprehensive, and understandable online health information (http://www.jmir.org/2018/2/e47/). Privacy, security and trust in the Patient Centered Infrastructure are still high priority first order requirements but become relative to the goals of the patient. As long as patients execute their rights on data privacy they are in charge to govern all data related processes. Part of the Patient Centered Infrastructure would be a possibility for patients to become more than only data subjects. GDPR and other regulation assume that data controllers are distinct entities that support patients rights. For a Patient Centered Infrastructure it makes sense to define data owners that also own part of the Patient Centered Infrastructure regarding the use of their own data. Even if that sounds a little constructed it makes sense if you think about the consequences of data policies supporting the wrong goals. There is enough evidence even if anecdotal that privacy kills and openness saves life (in some situations) (https://www.jmir.org/2019/11/e17045/). Information retrieval and access become even political dimensions in that sense. My personal guiding principle deduced from that is: Information cures (see upcoming post). This is why you won’t want patients not having this cure. Patients as information provider are then curing actors, health providers but only in a context that supports the right level of access and literacy: the Patient Centered Infrastructure

Education (see upcoming post) and communication reflecting the individual level of health literacy also requires a very flexible mechanism to query information and calculate results according to the current use case. Internet health information seeking can improve the patient-physician relationship depending on whether the patient discusses the information with the physician and on their prior relationship (http://www.jmir.org/2017/1/e9/). 

Lets collect some requirements as user stories regarding information retrieval from the Patient Centered Infrastructure (not exhaustive at all):

  • As a patient I want to be an active stakeholder in consumer health information technology development (again all stakeholders are required in that role) for the Patient Centered Infrastructure
  • As a patient I want constant participation in definition of requirements, interactions or processes for information retrieval and quality from the Patient Centered Infrastructure
  • As a patient I want to have and control trust in health information and belief in the effectiveness of information security for information management and retrieval in the Patient Centered Infrastructure
  • As a patient I want control all security levels of the data and information that I own
  • As a patient I want to understand the tools and methods I use for information retrieval from the Patient Centered Infrastructure
  • As a patient I want to select the tools and methods I use for information retrieval from the Patient Centered Infrastructure
  • As a patient I want to retrieve information that fits perfectly into my personal health pathway/process without the need for me to specify the informational need explicitly. (I just own my information and the process; the information need is determined implicitly but I can adjust it)
  • As a patient I want to issue queries and retrieve information according to my individual level of health literacy and my learning curve
  • As a patient I want to be an information provider in the Patient Centered Infrastructure with a transparent contribution for other stakeholders to use it for the health and wealth of others

How to do all that? 

According to the conception of this post I could have asked question and specified requirement only in relation to things that I personally understand. Which kind of means or gives at least hope that technology exists that could support all requirements and use cases. I personally do believe that we are not solving a technological problem here. Everything that is required could be leveraged by human interaction alone. Technology just makes it faster and at some points more reliable. Retrieval from a Patient Centered Infrastructure requires a cultural change to make accessible what already exists. 

And change is underway, accompanied by technology or driven by it. If it is a way to help patients to receive the right educational material about diabetes (https://www.jmir.org/2017/10/e342/) or a method to support information retrieval from web scale sources to evidence based practices (https://www.jmir.org/2019/8/e12621/) (even if targeted mainly at health care practitioners).

All this shows that technology only helps to ask the right questions but is not the principal reason for change. 

Stay tuned for upcoming posts about knowledge management and education…

Tags: , , , , , , , , ,
Posted in Education, General, Information, Privacy, Security and Trust, Web | Comments Off on A Patient Centered Infrastructure – Part 4 – Information Retrieval
Comments are closed.