How mHealth is changing Africa

Picture courtesy of CNN.com

DICOM and challenges it faces in the ever changing healthcare IT environment

In this writing I will explain the Digital Image Communications in Medicine (DICOM) standards. I will consider the current and proposed changes in informatics information infrastructure, the challenges they may have on DICOM standards and why.

DICOM is made up of a long list of standards. According to Merge Healthcare (n.d.), these standards are:

Conformance – These are statements written to allow a network administrator to plan or coordinate a network of DICOM applications.
Information Object Definitions and Service Class Specifications – This standard defines the types of services and information exchanged using DICOM.
Data Structures and Encoding & Data Dictionary – This is how commands and data should be encoded for interpretation.
Message Exchange – This is the standard by which two DICOM applications mutually agree upon the services they will perform over the network.
Network Communication Support for Message Exchange – How message will be exchanged using TCP/IP and OSI.
Common Media Storage Functions for Data Interchange – This is the DICOM model for storage of images on removable media.
Media Storage Application Profiles & Media Formats and Physical Media for Data Interchange – Specifications on physical storage media and details of the characteristics of various physical medium and media formats.
Grayscale Standard Display Function – Display function for display of grayscale images.
Security Profiles – Secure network transfers and secure media.
DICOM Content Mapping Resource – Defines templates and context groups used elsewhere in the standard.
Explanatory information – Consolidates informative information previously contained in other parts of the standard.
Web Access to DICOM Persistent Objects (WADO) – Specifies a web-based service for accessing and presenting DICOM persistent objects.

Proposed changes in the US health informatics infrastructure include the adoption of health information exchanges and this can present a problem with DICOM and multi-site use. According to Langer & Bartholmai (2011), one challenge of multi-site interoperability is “the magnitude of data produced by imaging systems and unstructured text reports.” They go on to say that this is because this giant amount of information “has made it very difficult to share results among sites until the wide availability of both broadband networks and universal protocols.” This is definitely and challenge to image transmission.

Another transmission challenge is the use of XML, which is utilized by HL7 V3.0. One example is caBIG, architecture (Langer & Bartholmai, 2011). Langer & Bartholmai mention that a “vast majority of PACS and imaging modalities will require intervening computers to broker the DICOM to caBIG translation.

Another challenge is storage. Steve Langer (2011) writes, “images from different vintage equipment will encompass different levels of the standard.” What this means is that there can be ambiguity in what data elements mean, based on time periods when it was created and the location of data elements if its not in a standard location (Langer, 2011).

There are several challenges that DICOM face. These deal with image transmission because of large data sets and lack of utilization between standards. Data storage can also be an issue because of legacy definitions.

References:

Langer, S. (2011 November 12). A Flexible Database Architecture for Mining DICOM Object: the DICOM Data Warehouse. Journal of Digital Imaging, 25, pp. 206-212. DOI: 10.1007/s10278-011-9434-6

Langer, S. & Bartholmai, B. (2011 February). Imaging Informatics: Challenges in Multi-site Imaging Trials. Journal of Digital Imaging, 24(1), pp. 151-159. DOI: 10.1007/s10278-010-9282-9

Merge Healthcare. (n.d.). The DICOM Standard. Retrieved from http://estore.merge.com/mergecom3/resources/dicom/java_doc/javadoc/doc-files/dicom_intro.html

Data Modeling: A Crash Course

The healthcare industry is making great steps towards using IT to provide better healthcare outcomes, increase patient safety and attempt to reduce costs.  Behind every one of these initiatives should be a well thought out IT system.  The backbones of these systems are databases.  But how does a HCO or software company start out creating this?  It all begins with system analysis and design and this cannot be done without data modelling.  Here is your crash course to data modelling and why it is important.

Data Modeling

In the data modelling process there are three items that must be completed for success.  These are Conceptual Data Models, Logical Data Models and Physical Data Models.  In this writing I will explain what data modelling is briefly, what each of these models are and why they are important to the development phase of a software system.  I will also be using an example of new patient check-in system for Dr. Model’s office that allows patients to enter demographic data upon arrival.

First, what is data modelling and why is it important?  Data modeling is, according to Margret Rouse (2010), “…the formalization and documentation of existing processes and events that occur during application software design and development.”  When data modelling analysts will use tools and techniques to translate the complexities of a system into easily understandable data-flows and processes.  These are used as the basis for construction of a database system or the re-engineering of one (Rouse, 2010).  The items that come out of this exercise show the data at various levels of granularity.  Also, by having well-documented models, stakeholders can eliminate errors before coding a new system (Rouse, 2010).

Conceptual Data Model (CDM)

This is the highest level of data modelling CDM is a vital first step when it comes to systems analysis and design.  Pete Stiglich (2010) explains that by creating a CDM  “key business entities or objects and their relationships are identified.”  For example at Dr. Model’s office, the current system of collecting information would be to let the patient fill out a form.  The designers now create a CDM, that shows the various entities that will be needed, such as a patient table, insurance company table, date table, and there could be others.  The model will then show which entities are related to one another.  So for example the patient table will have relationships to all the other tables.   It would be good to note that at this point this is a very high concept level showing what tables are related.  When creating the CDM, stakeholders will inevitably find many-to-many relationships, but these will be addressed in the logical data model (Stiglich, 2008).

Logical Data Model (LDM)

1keydata (n.d.) defines a LDM as a model that “describes the data in as much detail as possible, without regard to how they will be physical implemented in the database.”  In this model analysts will now start to identify the fields that will be in the tables and the relationships between them (1keydata, n.d.).  They will also identify primary & foreign keys, and begin the process of normalization.  As a side note normalization, as defined by Mike Chapple (n.d.), is “the process of efficiently organizing data in a database.”  Normalization is what will address the many-to-many relationships, and get rid of redundancies.  Going back to the application for Dr. Model, the analysts now include fields in the tables.  For example, the patient table will include patient first & last name, medical record number (primary key), patient sex, patient age and many other bits of demographic information.  

Physical Data Model (PDM)

Exforsys Inc. (2007) defines this data model as “the design of data while also taking into account both the constraints and facilities of a particular database management system.”  This is where analysts take the LDM and make sure that all the pieces of data are configured based on the environment.  To further explain, as in our Dr. Model case and the patient table, now we define the patient first name as a variable character type field (varchar) and state how many characters it can contain.  The age field will be defined as an integer data type.  Knowing this information from the PDM can be useful in estimating storage needs (Exforsys Inc, 2007).  Analysts must also take note that this model may be different based on the type of database management system that will be used (Oracle, MS SQL Server, MySQL, etc.) (1keydata, n.d.).

In conclusion the end result of an application is reliant on a strong data modelling process.  This process will build the foundation for the database and without it many issues can occur.  It starts off as a very high level CDM, fills in the blanks with the LDM and then makes sure that it will all fit in a neat package with the PDM.  Any HCO needs to rely on this modelling process so that it can maintain quality data for the foreseeable future.

References:

1keydata.  (n.d.).  Logical Data Model.  Retrieved from http://www.1keydata.com/datawarehousing/logical-data-model.html

1keydata.  (n.d.).  Physical Data Model.  Retrieved from http://www.1keydata.com/datawarehousing/physical-data-model.html

Chapple, M.  (n.d.).  Database Normalization Basics.  About.com.  Retrieved from http://databases.about.com/od/specificproducts/a/normalization.htm

Exforsys Inc.  (2007 April 23).  Physical Data Models.  Retrieved from http://www.exforsys.com/tutorials/data-modeling/physical-data-models.html

Rouse, M.  (2010 August).  Data modeling.  SearchDataManagement.  Retrieved from http://searchdatamanagement.techtarget.com/definition/data-modeling

Stiglich, P.  (2008 November).  So You Think You Don’t Need A Conceptual Data Model.  EIMInstitute.org, 2(7).  Retrieved from http://www.eiminstitute.org/library/eimi-archives/volume-2-issue-7-november-2008-edition/so-you-think-you-don2019t-need-a-conceptual-data-model

How important are metadata and data dictionaries?

How important are metadata and data dictionaries?

In data warehousing and data storage, metadata and the use of a data dictionary is extremely important.  Metadata in layman’s terms is basically data about data.  It explains how the data was created, when it was created and the type of data it is.  Staudt, Vaduva & Vetterli (n.d.) state in their paper that when working with the complexity of building, using and maintaining a data warehouse, metadata is indispensible, because it is used by other components or even directly by humans to achieve particular tasks (Staudt, Vaduva & Vetterli, n.d.).  

Metadata can be used in three different ways:

• Passively – documents the structure, development process and use of the data warehouse system.  (Staudt, Vaduva & Vetterli, n.d.)
• Actively – Used in data warehouse processes that are “metadata driven.”  (Staudt, Vaduva & Vetterli, n.d.)
• Semi-actively – Stored as static information to be read by other software components.  (Staudt, Vaduva & Vetterli, n.d.)

So as you can see, the use of metadata is important, not only to store information about the data, but it is also used in processes by the data warehouse and by other applications.  Metadata also improves on data quality by providing consistency, completeness, accuracy, timeliness and precision.  This is because it provides information on the creation time, and author of the data, the source and the meaning of the data when it was created. (Staudt, Vaduva & Vetterli, n.d.).  

Regarding the data dictionary, this reminds me of when I would query the database at a past position of mine.  Because there was no data dictionary, it was hard to manually decipher the relevance of the data that I was searching for and where it was stored.  Because of this, I did not always bring back the correct fields necessary to complete my work and this caused devalued use of time.  On AHIMA’s website Clark, Demster & Solberg (2012) prepared an article about the use of a data dictionaries and how they can be used to improve data quality. 

• Avoid inconsistent naming conventions
• Avoid inconsistent definitions
• Avoid varying lengths of fields
• Avoid varied element values (Clark, Demster & Solberg, 2012)

By using the data dictionary, there is a consistency created in the data, which in turn improves data quality.

In conclusion, both metadata and data dictionaries are vital to creating consistent data.  This data can be tracked and can be used to create interoperable processes between the data warehouse and other applications.  Without these, architects are taking a chance and increasing their opportunities for use of less quality data.

References:

AHIMA. (2012 January).  Managing a Data Dictionary. Journal of AHIMA 83(1),pp. 48-52.  Retrieved from http://library.ahima.org/xpedio/groups/public/documents/ahima/bok1_049331.hcsp?dDocName=bok1_049331

Staudt, M., Vaduva, A. & Vetterli, T.  (n.d.).  The Role of Metadata for Data Warehousing.  Retrieved from http://www.informatik.uni-jena.de/dbis/lehre/ss2005/sem_dwh/lit/SVV99.pdf

eHealth, Patient-Interaction and Data Quality

eHealth, Patient-Interaction and Data Quality

Facebook, Instagram, Twitter, Pinterest.  Have you ever heard of these?  Of course you have.  Over the last several years, consumer interaction with computers has soared.  People are now shopping, video chatting, expressing themselves, posting pictures and conducting business using computers, and this is just the tip of the iceberg in terms of how the computer is being used to interact with the world.  And the computer is not the only means.  Using smartphones, the masses can now do all of this on the go, creating a never-ending flow of information.  What is this information made of? Data!

It is only natural that the healthcare industry, now venturing into the IT world, would want to jump on the train to ride the information highway.  With all the human computer interactions going on, and in the lay community, it only makes sense to find a way to have patients interact with their own health data.

Hesse & Shneiderman (2007) wrote the following in the abstract of their article.

“New advances in eHealth are prompting developers to ask “what can people do?” How can eHealth take part in national goals for healthcare reform to empower relationships between healthcare professionals and patients, healthcare teams and families, and hospitals and communities to improve health equitably throughout the population?” (Hesse & Shneiderman, 2007)

The data that is used by patients and providers goes beyond the one-to-one relationship between them.  This is the value that the emerging area of eHealth brings.  The physician is a “microunit” of a larger systems, which includes the care delivery team (nurses, office staff, etc) and the patient is also a “microunit”, surrounded by family & friends, and the community (Hesse & Shneiderman, 2007).

One way that patients can help improve quality of data is by looking up information on their own, and through a provider-sponsored portal (Geissbühler, 2012).  This will help them in “assisting” the provider with pinpointing their issues, while having a reliable source of information. 

Another emerging trend are “patient-controlled health information exchanges” (Geissbühler, 2012).  These are ways that a patient can control access to their health information, federating documents from several providers or organizations and even provide their own contributions (Geissbühler, 2012). 

A third way patients can interface with their health information is by using their “’digital proxy’, a mobile, always-on, permanently connected, and context-aware device such as a smartphone.” (Geissbühler, 2012).  Also homes can be made intelligent, and they can be made aware of the needs of their inhabitants (Geissbühler, 2012). 

Some of this may seem science fiction, but this is the way the health industry is moving.  People are getting more comfortable with sharing information online in the social networking world.  The use of smartphones is proliferating and can be a valuable asset.  All of this information sharing, and allowing patients to control their own data will help to increase data quality on a whole.  This is good not only for the patient, but also for the providers and the communities that the patients live in.

References:

Geissbühler, A.  (2012 June 2012).  eHealth: easing the transition in healthcare.  Swiss Medical Weekly, 142.  doi:10.4414/smw.2012.13599

Hesse, B. W. & Shneiderman, B.  (2007 May).  eHealth research from the user’s perspective.  American Journal of Preventive Medicine, 32(5), pp. S97 – 103.  DOI: 10.1016/j.amepre.2007.01.019

Keeping Data Safe

In order to protect information systems and data, a best practice is for organizations to develop and maintain a data security program.  Examine the essential elements of a health care information security program and why each element identified is essential.  Be sure to examine both technology and human factors. 

Columbia, SC, October 30, 2012, “As many as 657,000 S.C. businesses had their tax information stolen in the massive security breach at the state Department of Revenue…” (Shain, A., 2012).  October 2012, “Hackers were able to breach more than 60 Barnes & Noble (BKS) stores, including locations in New York City, Miami, San Diego and Chicago, and obtain credit card information…” (Graziano, D., 2012).  Again in October of 2012 a Vermont credit union accidently threw away two backup tapes, which could affect up to 85,000 individuals (Walker, D., 2012).  October of 2012 was a busy month for data breaches.  The breaches highlighted above have nothing to do with the health care industry, but as the proliferation of EMRs, EHRs, PHRs, mHealth, wireless technology, electronic claims processing and HIEs continue, HCOs will need to remain vigilant in protecting PHI. 

In 2009 Clifton Phua, wrote an article about computer fraud and security.  In it he noted that 81% of security breaches were from malicious outsiders, 17% from malicious insiders and 2% from unintentional insiders (Phua, C., 2009).  This means that anyone housing sensitive data must take measures to lock those who would want to get into the systems out, and to make sure those who have appropriate access do not intentionally or unintentionally disseminate protected information.

There are several ways to assist in ensuring that data will not be breached.  The first is on the technology side.  These include firewalls, intrusion detectors and robust anti-virus protection (Phua, C., 2009).  Firewalls stop intruders from getting into your private network through security rules and other measures.  There are two types, software firewalls and hardware firewalls.  If by chance someone does get in an intrusion detector will send alerts and appropriate measures can be taken.  Lastly anti-virus will stop malicious software from getting on the network, creating back doors for intruders.

On the human side one, HCOs should implement data handling policies (Phua, C., 2009).  Some items in a policy like this could be:

·      Masking unneeded data, such as the first digits of a social security number
·      Shredding paper and physically destroying hard drives
·      Reviewing the what, where and how’s of data
·      Background checks on employees
·      Auditing employees access to data
·      Data encryption on laptops and other portable devices, to protect information in the case of theft (Phua, C., 2009)
·      Using strong passwords

And the list goes on.

Another check that can be put in place is the use of thin clients.  These are machines that access data in a client-server fashion.  The machines can also have copy and paste, USB drives and other ways to export data disabled (Phua, C., 2009).

In conclusion, when measures are taken to protect PHI, both the use of technology and ways to recuperate from human error must be taken into account.  Attacks can come from the inside as well as the outside and breaches can be intentional and unintentional.  A wise plan takes all of this into account and sets up a roadmap towards a more secure infrastructure.

References:

Graziano, D.  (2012 Oct 24).  HACKERS STEAL CREDIT CARD INFORMATION FROM 63 BARNES & NOBLE STORES.  Retrieved from http://bgr.com/2012/10/24/barnes-noble-security-breach-credit-card-information/

Phua, C.  (2009 Jan 01).  Protecting organisations from personal data breaches.  Computer fraud & security, 2009(1), 13 - 18.  DOI: 10.1016/S1361-3723(09)70011-9

Shain, A.  (2012 Nov 01).  Data security breach expands to 657,000 S.C. businesses.  Retrieved from http://www.mcclatchydc.com/2012/11/01/173313/data-security-breach-expands-to.html

Walker, D.  (2012 Oct 26).  Vermont credit union discards unencrypted data of 85,000.  Retrieved from http://www.scmagazine.com/vermont-credit-union-discards-unencrypted-data-of-85000/article/265522/