EHRs Pave The Road for New Achievements in Health IT

Linguistic tools can solve the problems that users of EHRs face


The global business environment is becoming increasingly digital thanks to the high overall level of technological development. The health system does not stay behind, as well. Medical records are now made in the electronic form in many countries, and the growing amount of data offers unprecedented opportunities for secondary use of this data for research. Electronic health records (EHRs) can be used in the population and disease research, the detection of ADRs, improving the overall quality of care, and more.

In the U.S., the first meaningful step towards implementation of EHRs was made less than 10 years ago due to generous provider incentives for those who demonstrates 'meaningful use' of EHRs. These initiatives were included in the HITECH ACT. For instance, 12.2% of non-federal acute care hospitals had at least a basic EHR in 2009, and nearly all such hospitals possessed a certified EHR technology in 2014.

The recent study showed that adoption of electronic medical records is high in many countries, such as Italy, Saudi Arabia, the UAE, and more. The adoption is not very high in South Africa and Brazil, but they are expected to catch up to the leaders in future.

There are many EHR solutions that can be used by medical institutions. Some of the most popular are developed by McKesson, eClinicalWorks, Nuesoft, Practice Fusion, and other developers of medical data analysis software . The solutions have many common features: appointment management, patient portal, e-prescribing, etc. Yet they have some unique features like a single patient facesheet, a clearing house, an ability to set five personal menu objectives and measure them in real time, to name a few.

Doctors and EHRs

But not all doctors are quick on the uptake of EHRs. Some of them cannot get used to a new system for a long time given the difference in rates of learning and adaptation, and this results in significant losses in productivity. Many doctors still prefer to make voice records like they have been doing for years. However, it is sometimes very hard to find the glean the required information (e.g. tests that were done). Let's discuss these issues in detail.

Doctors Still Prefer Recordings

Audio records still remain a preferred choice for many doctors. First of all, this is an approach used for many decades. The doctor may have had several unsuccessful attempts to start using EHRs and does not want to try again. Secondly, the older a doctor is, the higher the chances that he/she never starts using such a system. Another important reason is that doctors face time constraints. The ease of dictation enables doctors to include important details on the patient's history, examination, discussions of a case with other doctors, etc.

Audio records can be included in EHRs, though. Doctors can listen to their records themselves and enter the data into the EHR. The quality of transcription will be high in this case, but doctors do not have enough time for that in practice. The records can also be sent to the transcription department or a medical transcription company to be transcribed. How good will the quality of the transcription be in these cases? The recording may contain some really difficult words (e.g. chlorobenzylidenemalononitrile) that may be difficult to decipher. So the quality of the report may suffer if a medical transcriptionist who works with the recording is not skilled enough. Taking into account the fact that many hospitals use transcription companies based abroad (due to a lower cost), mistakes may occur in the report simply due to non-native speakers needing to interpret unclear phrases.

What can be done to overcome this problem? Voice recognition software can be used for this purpose due to their high level of accuracy (up to 90%). There are many speech recognition solutions designed for clinicians like Dragon NaturallySpeaking and SMARTMD. Some EHR systems even have built-in voice recognition solutions that enable doctors to directly narrate to the system (e.g. Winscribe Text).

Is Structured Data Good Enough?

Unlike written notes which are in the form of free text which is difficult to analyze and structure by computers, data stored in EHRs is believed to be well-structured. But is it easy to find the required information in such a system? There are some issues every person who uses an EHR may face. The systems cannot boast a good balance between free text and structured data. EHRs are mainly 'template-oriented', and users have to choose one of the forms they are suggested by the program to enter the data. These forms are not flexible, and some important nuances like contextual information and the doctor’s 'if and then' statements remain untapped. So the clinical picture may be incomplete.

Information search is also a serious problem. It is a common situation when a clinician is frustrated with the time taken to find critical elements of clinical importance in EHR forms. Some key details can still be missed since EHR records are hard to read and lack good usability. Spelling and lexical errors may occur and a user may not find the required data by typing keywords in a search box (if there is any). Many EHRs have the copy-and-paste function and users copy and paste information about the patient from an older record to the new one. This leads to the so-called 'note bloat' and even mistakes. For instance, one hospital was sued by the patient who did not get proper treatment as attending physicians did not notice some critical details in the medical history.

Linguistic Approach to the Use of EHRs

Linguistic tools like Intellexer can be used to deal with issues related to the use of EHRs. For instance, Summarizer can be used to get a brief summary about a patient eliminating the need to read all the records available in the system. The summary will the most relevant information which will be theme-oriented and concept-oriented. This means that the sentences will be mostly relevant to the specified topic and their importance will be determined with respect to a number of user-defined concepts.

Spelling errors can be automatically corrected by Spellchecker and they will not impact the search results. There is one more problem that was not mentioned above: the data exchange between EHRs. Though EHR adoption is high, there is still a lack of interoperability between EHRs, even if they were developed by the same company. The fact is that there is no standard solution that could cater to the needs of every medical organization. EHRs are often customized after they are purchased, creating obstacles to the information exchange. This makes it difficult to exchange information between two organizations with similar, but different systems. 

How can linguistic tools help? Let's imagine that there is access to data stored in different EHRs and there is a need to find differences in the clinical history of the same patient. Comparator can give clinicians a helping hand in such a case. This solution can accurately compare the documents and set the degree of similarity between them.


Now is the age of rapidly evolving health IT and EHRs represent one of the most important developments. EHRs can be used not only to improve the quality of healthcare but for different research purposes. Yet people using EHRs face some difficulties that can be overcome through technology, such as linguistic solutions. 

Big data hype small

Read next:

Is Big Data Still Overhyped?