Accent on Objects

It has been many years since I acknowledged patient record subpoenas for medical malpractice lawsuits and other legal actions as an HIM professional and designated custodian of records (COR). But the process was memorable.

During the 1970s, one was not able to reproduce analog paper and photographic film or send records by postal mail or courier to the courts.  Rudimentary paper and film photocopy machines only recently were introduced into healthcare organizations, and the courts required the personal delivery of “original” source documents and records by a COR.  Consequently, upon receiving patient record subpoenas, I took a large cardboard box and collected from each department the “original” source documents required by the subpoenas. The contents included the patient’s paper financial and medical records.  The medical records also included all film-based diagnostic images, tape-based medical dictation, cine-based ECGs, and pathology slides.

During the 1980s, when I established my related career in HIT and because of my COR experiences during the “analog” years, I knew well that electronic patient records consisted of more than just the structured data typically found in electronic patient financial and medical records. (Structured data are the record’s binary, discrete, and computer-readable data elements that, typically, are stored in relational databases with predefined fields.)  Electronic medical records (EMRs) also consisted of digital diagnostic images, audio file-based dictation, and ECG waveforms. In fact, such unstructured data make up at least 75% of all the data in a typical patient’s EMR.  (Unstructured data are the record’s non-binary, non-discrete, and often human-readable data elements that, typically, are contained in text-based reports, emails, and web pages and include symbols, images, video clips and audio clips.  In some vertical markets, unstructured data are referred to as a record’s intellectual substance or content.  In technical arenas, unstructured data are referred to as “objects”.)

Just like healthcare organizations, the courts finally have entered the digital age. Today, secured electronic files of “original,” electronic source documents and records as well as “copies” of original, electronic source documents and records are admissible in courts as long as the healthcare organization can substantiate (1) the trustworthiness of the system(s) used to store and retrieve the documents and records; (2) the accuracy of the organization’s records management policies and procedures; and (3) the documents and records were not created (or altered!) just for a court case. (NOTE:  Always one must verify the courts’ acceptance of digital records on a state-by-state basis.)

Large cardboard boxes have been replaced by EMR (or other system) features that promote single points of personalized access through which to find and deliver electronic information, applications, and services. As such, in either hybrid or full EMR environments, designated CORs, Release of Information professionals, and even patients—after rigorous authorization and authentication processes—merely click on hyperlinks and instantaneously retrieve “original” electronic source documents and records required by subpoenas or other requesters.

While our industry continues to pursue the best “highways” to securely transmit the documents to and acknowledge receipt from requesters, today’s day-to-day challenges involve the current mechanisms used to transmit unstructured data and the shameful output of structured data generated by most EMR systems.

For example, the transmission of the large and ever-growing number of patient diagnostic images (primarily radiology images), which remain hand-carried or sent by postal mail or courier from hospitals, physicians / groups, specialty (e.g., cancer) centers, etc., to other hospitals, physicians / groups, and specialty centers on CD storage media, is completely unmanageable. Many of the CDs containing (e.g., radiology) diagnostic images cannot be imported into the receiving Radiology PACS due to the way the images were burned into the CDs.  Although most of the CDs include the senders’ viewers for measuring, window / leveling, etc., often the CD files arrive corrupted.  Frequently the CDs are misfiled and / or lost.  Consequently, transmitting diagnostic images on CDs has lead to duplicate testing with more patient exposure to radiation.  In addition, when the CDs contain diagnostic images other than radiology images, often the receivers have no corresponding PACS for these other, “ology” images.

Thankfully, popular, standard, inbound (i.e., CD ingestion and electronic receipt of diagnostic images) and outbound (i.e., report and image distribution to referring physicians, referral centers, etc.) image sharing solutions exist.  However, most are too expensive for the healthcare provider masses.  In addition, few, if any, non-standard image sharing solutions exist, whereby direct connections are established between two or more organizations for readings, consultations, and 2nd opinions and inbound and outbound electronic reports accompany the images.

Also, there is not a healthcare professional that has not experienced the reams of paper output generated by EMR systems because the systems’ structured data are not report-formatted for output. This is one reason why a patient still cannot receive his/her entire patient record from a portal!  Not that I promote hard copy printing; however, healthcare providers still must maintain a legal archive from which to generate the electronic document presentation as proof for exception and dispute handling.  In other words, providers must have the document presentation for legal purposes and NOT an informational statement or data representation of the document, which, unfortunately, remains common in today’s electronic patient record system output.

Cleaning Up Document Imaging’s Image

Since the late 1980s, when one of the technologies of electronic document management systems (EDMSs), document imaging technology (a.k.a., document capture technology), first was implemented in U.S. healthcare provider organizations to manage (i.e., digitize) the organizations’ massive amounts of medical, financial, and administrative analog documents, many health information management (HIM), healthcare information technology, and other related professionals did not understand that document capture technology is NOT the only technology of an EDMS.   Other EDMS technologies include:

Automatic IdentificationTechnologies – EDMS technologies that allow documents to be automatically identified.  The following are examples of EDMS automatic identification technologies:

  • Automatic / Intelligent Document Recognition (ADR/ IDR) recognizes the layout and content of generic form types.  In healthcare organization EDMSs, ADR / IDR is used to recognize digital and analog form categories.
  • Bar codes are machine-readable representations of data, typically dark ink on a light background to create high and low reflectance which is converted to 1s and 0s. Originally, bar codes stored data in the widths and spaces of printed parallel lines.  Today, bar codes also store data in patterns of dots, concentric circles, and text codes hidden within images. Bar codes are read by optical scanners or scanned from an image by special software.  In HIM department EDMSs, bar codes are used to eliminate the manual indexing of document type, patient name, provider name, and medical record number, etc., as well as medical record “separator” sheets during the digital scanning process.
  • Intelligent Character Recognition (ICR) allows different styles of hand printing and hand writing to be learned by a computer.   Most ICR software includes a self-learning “engine” (i.e., a neural network) that automatically updates the recognition database for new handwriting patterns.  In healthcare organization EDMSs, ICR is used to recognize documents’ hand printed / hand written numbers or abbreviations on analog orders, in analog progress notes, etc.
  • Optical Character Recognition (OCR) is the electronic translation of analog typewritten or printed text into machine-editable text.   Early OCR systems required system training to read specific typed or printed fonts.  Today, OCR has a high degree of recognition accuracy for all fonts.   Some systems are capable of reproducing formatted output that closely approximates the original scanned page including images, columns and other non-textual components.  In healthcare organization EDMSs, OCR is used primarily to recognize the printed text on claim forms.
  • Optical Mark Recognition (OMR or Mark Sense) is the process of capturing data by requiring a page image to have high contrast and an easily-recognizable shape.   Mark sense (OMR) is distinguished from OCR by the fact that a recognition engine is not required.  That is, the marks on a page are constructed in such a way that there is little chance of not reading the marks correctly.   One of the most familiar applications of optical mark recognition is the use of #2 pencils on paper-based, multiple choice-type question examinations.  In healthcare organization EDMSs, mark sense often is used for physician office analog super bills and charge tickets.

Enterprise Report Management (ERM) technology (formerly known as COLD: Computer Output to Laser Disk) – an EDMS technology that stores “computer output” to – and indexes “computer output” on – digital storage media.  Once stored, the computer output can be easily retrieved, viewed, printed, or distributed.  Today, “computer output” consists primarily of batch-generated computer reports with data that are report-formatted.  Electronic Health Record (EHR) batch-generated computer reports include EHR system bills / invoices and management reports.

Document Management (DM) technologies – EDMS technologies that control and organize documents.  The following are examples of EDMS DM technologies:

  • Document Assembly allows documents to be automatically retrieved in the “correct” order, based on pre-defined, user-specific rules and tables.
  • Document Version Control allows documents to be automatically assigned version numbers.  For example, this includes daily laboratory test result reports (version 1) vs. cumulative summary laboratory test result reports (version 2); preliminary radiology procedure result reports – unsigned (version 1) vs. final radiology procedure result reports – signed (version 2); transcribed operative reports (versions 1, 2) vs. signed transcribed operative reports (version 3) vs. amended transcribed operative reports (version 4).  Typically, only the most recent or last document version is accessible for view purposes.
  • Document Check-in / Check-out allows users to collaboratively review / edit shared documents without concern about who might be simultaneously updating the document and to view all the entries made to the shared document.  Clinical teams that author electronic progress notes is an example of this important DM technology.
  • Document Security consists of all the technical document tools to protect, control and monitor document access (e.g., unique user identification / authentication, audit trails, automatic log-offs, and biometric identifiers) to prevent unauthorized access to documents transmitted over a network.

Digital signature management technology – an EDMS technology that offers both signer and document authentication. Signer authentication is the ability to identify the person who digitally signed the document. The implementation of the technology is such that any unauthorized person will not be able use the digital signature. Document authentication ensures that the document and the signature cannot be altered (unless by means of showing both the original document and the changed document). As such, document authentication prevents the document signer from repudiating that fact.

Forms Processing (a.k.a., Electronic Forms Management, Automated Forms) technology  – an EDMS technology that electronically delivers paper forms for printing and completion, accepts scanned paper forms and extracts data from the boxes and lines on the forms to populate databases, and utilizes eForm templates — that look like paper forms — for online data entry / data collection.   Because patient medical records consist of hundreds of forms, even with EHRs, this EDMS component technology is important for healthcare organizations.

Workflow / Business Process Management (BPM) technology – an EDMS (or any other type of information system) technology that is the EDMS’ (or other information system’s) most important component technology!   BPM technology automates business processes, in whole or in part.    EDMS workflow / BPM technology passes documents, information, or tasks from one user to another for action according to a set of business rules.

Today, many healthcare organizations use the EDMS’ document capture component technology more than the other EDMS component technologies.  The primary reasons include the need to:

  • continue to convert existing analog documents into digitized documents
  • maintain a legal archive from which to generate the electronic document presentation as proof for exception and dispute handling (i.e., by extension, the need to have the document presentation for legal purposes and NOT an informational statement or data representation of the document, which, unfortunately, remains too common the output in today’s electronic patient record systems).

However, when ALL internal and external documents, forms, notes, letters, reports, and messages are digitally-generated, stored, and distributed, the use of the EDMS document capture component technology will be drastically reduced, perhaps eliminated, and all the other EDMS component technologies will be used frequently.

The Cost Savings Realized by Single-Platform Solutions

Medicare cutbacks, increased mergers and acquisitions, and declining revenues and margins are daily realities for healthcare provider organizations. Yet so are costly and often mandatory IT initiatives, such as replacing or implementing electronic health and financial record (EH&FR) systems, integrating and securing the next mobile form factor of computing devices, converting to ICD-10 for reimbursement purposes, and meeting the Affordable Care Act requirements.

How does a CIO not compromise on health information technology (HIT) delivery and still maintain cost containment priorities?

A well-developed, single HIT platform strategy for most applications can realize cost savings by optimizing the IT infrastructure, regardless of whether the platform resides on premise or in the cloud. For purposes of this article, a single computing platform is defined as one that provides a database and set of technologies across a set of heterogeneous applications.

However, still many provider organizations acquire and deploy heterogeneous applications that could be built on a single, common platform. Consequently, this approach continues to generate silos of systems that cannot interoperate and to burn holes in IT operating expense budgets.

Thankfully, no longer is the HIT industry faced with an abundance of best of breed EH&FR applications, typically with each sitting on top of its own proprietary computing platform. Top-selling EH&FR solution vendors finally are offering one or — at most, two — unified platforms for the myriad, structured data-based EH&FR applications.

On the other hand, today, many solutions that complement EH&FR systems remain siloed. Such solutions worth mentioning for this article include but are not limited to Picture Archiving and Communication Systems (PACS) solutions for the storage of image-generating “ology” applications and Natural Language Processing (NLP) solutions for applications such as speech recognition, computer-assisted coding (CAC), and business intelligence (BI). By deploying a single storage platform for PACS solutions or a single NLP platform for speech, CAC and BI applications, chances are that the platform will be around longer than any application an organization has today.

For example, when developing a strategy for managing all the image-generating “ology” applications in the enterprise, it behooves the provider organization to consider a vendor-neutral archive (VNA). This poor choice of name for a storage platform (a more appropriate name might be a PACS-neutral or image file-neutral archive) stores all the heterogeneous, enterprise modality images (DICOM and non-DICOM) in a single, replicated archive. This eliminates the need for each costly, siloed PACS archive and, more appropriately, changes the “A” In PACS from “Archiving” to “Accessing.”

Typically, the platform includes tools to help manage all the different lifecycles of the image data (e.g., retention scheduling and purging mechanisms). And, perhaps the biggest benefit to clinicians is that this platform allows for the creation of an aggregated, patient-centric record of all the enterprise images.

A VNA platform is expensive, requiring a large, complex, capital-intensive project. But well thought out strategies for deploying this platform over time have proven that significant cost savings are realized. Eliminating the organization’s most costly departmental PACS operating expense — recurring migrations with storage media migrations typically occurring every three to four years and PACS replacements requiring massive proprietary data migrations, typically occurring every five to seven years — alone cost justifies this single platform solution.

When developing a strategy for deploying diverse yet complementary EH&FR applications such as speech recognition, CAC software, and BI, it behooves the provider organization to leverage a single NLP platform product that manages each of these technologies well. All NLP engines perform some type of pattern matching, and  to varying degrees, incorporate both rules-based and statistics-based techniques. That means that a plethora of applications can be developed using a single NLP platform.

These applications include but are not limited to speech to text (dictation systems), text to code (medical coding), data analytics (data warehouse systems), foreign language translation (Google Translate), question and answer (IBM’s Watson playing Jeopardy), document classification (Outlook’s spam filtering), candidate identification for clinical trials (determining eligible candidates), adverse events (patients with gunshot wounds), core measures (patients who already had been treated), geographic disease epidemiology, and on and on.

At HIMSS13, several complementary EH&FR application vendors were demonstrating more than one heterogeneous application using a single NLP platform. Currently the challenge is that not one vendor offers most of the above applications or even the same applications. Where one vendor might offer medical coding and data analytics using a single NLP platform, another vendor might offer data analytics and candidate identification using a single NLP platform.

For cost justification and interoperability purposes, it behooves provider organizations to strategically look under the hood and determine where a single NLP platform can be deployed.

Originally published at:

Structured and Unstructured Data, I Adore You Both

Calling all electronic patient record systems (EPRS) structured data! Yes, all you electronic health / medical, administrative, and financial systems’ data elements that are binary, discrete, computer-readable, and, typically, are stored in relational databases with predefined fields … you tidy, typically core, transactional and mined elements. Hello? I’m talking about all you digital, patient demographic, financial, and clinical health data that are sitting in master patient indices, insurance claims, clinical histories, problem lists, orders, test results, care plans, and business intelligence reports — to mention just a few.

Meet unstructured data! Yes, all you EPRS data that are non-binary, non-discrete, sometimes only human-readable, and sometimes not stored in relational databases. This means all you digital, bit-mapped images, text, videos, audios, and vector graphics that are harnessed in word-processed summary reports, electronic forms, diagnostic radiology images, scanned document images, electrocardiograms, medical devices, and web pages – again, to mention just a few.

I know this might be an awkward introduction. However, I’m really happy to finally get you two data formats together. And, while this might be jumping the gun a bit, I really hope one day you two will get married! I know, I know. That is, after you’ve carefully sorted out all your differences and learned how to live together in peace and harmony for the betterment of patient care.

After all, I’m certain you heard the rumor that the “adoption” and “Meaningful Use” of “certified” diagnostic image-generation and management systems, such as a PACS for one or more of the “ologies”, might be included in Stage 2. In addition, heaven help if, given the revised Federal Rules of Civil Procedure Governing Electronic Discovery that became effective December 1, 2006, a patient’s electronic health/medical, administrative and financial episode-of-care records (I mean x-rays, bills, ECGs, orders, progress notes – the works!) are subpoenaed for that Weird News Andy case we recently read on Mr.HIStalk! So, don’t you think it’s time at least to begin acknowledging one another in public?

Who am I, you ask, to be so bold to introduce you to the other? I’m just one, frustrated HIT professional who specializes in most of the EPRS unstructured data and who observes that these data are rarely considered in EPRS strategies and purchases … until after the fact. Once considered, they divide provider organization departments right down the middle; those working with you, structured data, vs. those working with you, unstructured data. Don’t even get me started about integration and usability issues!

Come close, structured data, so I can tell you that I do adore you – especially when I search a database for one or more of you, and, quickly and easily, the search engine finds, retrieves, and even manipulates parts or all of you. On the other hand, what often makes me want to delete you is when you insist on snubbing unstructured data. I’ve even watched you try to convert some unstructured data, such as rich-text or video data, to your popular religion, using pretty-good-but-not-perfect artificial intelligence and recognition tools … just so that you can brag about how you were able to generate the complete health story with your qualities.

Unstructured data? After so many years working with you, you know that I love being able to retrieve your gorgeous, bit-mapped, raster images generated by that digital chest x-ray or computed tomography (CT) scan stored in a diagnostic image management system; or, listen over and over to your brilliant sound bytes generated by that digital stethoscope; or, fast forward your streaming videos / frames generated by that important cardiac catheterization study; or, admire the perfect lines connecting the series of points plotted by that fetal trace recording. On the other hand, what I can’t tolerate is when I am required to search, for example, a valuable narrative text for one or more of you, and after hours I still can’t find you!

Today there is no complete electronic patient health / medical, administrative or financial record system without both of you. Let me see a hand shake.

(Originally posted at: