You’ve heard it called dozens of names. Voice transcription software, voice to text software, voice activated software or more commonly–voice recognition software. No matter which term you use to describe it, it’s all the same thing. But how is it used for medical transcription? And is it accurate enough for your practice?
“Speech recognition” can be defined as “translating words you speak into decipherable text.” The word “decipherable” is of key importance here. That is, because what the voice recognition software produces has to make sense and be true to what was actually said.
Voice recognition software used for medical transcription, though, depends on the software being able to recognize sound patterns, which are then paired with the right phonetic structure (or interpretation thereof) in established databases.
It’s not too difficult to imagine how many things can go wrong with this sound-to-written-text paradigm.
The Trouble With Voice Recognition Software in Medical Transcription
Voice recognition software (VRS) for medical transcription, for instance, has to make a number of arbitrary interpretations that artificial intelligence (AI) may not yet be fully capable to tackle perfectly.
For example, in order for medical transcription to be effective, it must be 100% accurate. In other words, a physician’s specific speech–such as nuances, accents and most importantly, intent–needs to be accurately picked up on by the VRS. These are things that, based on the technology we have today, only humans have the ability to pick up on.
For the most part, VRS for medical transcription can handle the phonetics-to-written-text part, as long as the speaker provides clear, audible, consistent and phonetically uninhibited text. In short, it allows you to type with your voice. The problem, though, is that human beings don’t always speak audibly, clearly and uninhibitedly. After all, we’re only human–right?
Deficiencies of Voice Recognition Software for Medical Transcription
The following are the most egregious shortcomings and limitations of speech recognition software technology:
VRS is too slow, especially in high-paced settings
If you’ve ever worked with voice recognition technology, you know that it requires lots of operating memory and up-to-date hardware in good operating condition. In general, most people can speak faster than they can type.
However, this isn’t to say that VRS is necessarily quicker, less-work-intensive and easier than typing information into a computer. Because that just isn’t true. At least not until the technology has improved on several levels. The extensive editing required by itself can easily offset any time supposedly saved by VRS.
VRS can’t always connect the dots
It takes children years to make connections between letters, words and thousands of different nuances—why should we assume VRS technology is any different? At its core, VRS is a computer. A technology created to document exactly what it hears. And it does just that–it records exactly what it hears. Which is actually a really big issue, especially for an industry like healthcare.
VRS can experience problems recognizing things like brand names, uncommon surnames and nouns. Usually these are words subjected to inflections–such as using the possessive form.
Such words may have to be altered later on by the physician or medical personnel. To make matters worse, persons with heavy or unusual accents may experience significant and sometimes unworkable problems with this technology. Issues also arise if the device if the microphone has a weak signal or if the speaker isn’t speaking clearly. The list goes on.
Inability to spot errors
VRS for medical transcription cannot easily spot human errors such as inconsistencies in context that human can easily detect.
Even simple things like words that sound alike can present a huge problem for voice recognition software. Humans can usually decipher which sound-alike words belong in a sentence by knowing the overall meaning of the sentence. This includes the tone of the speaker and the context of the speech.
Example: Knowing when to use the words “two” or “to” instead of “too”.
The type of speech needed for accurate results
Not only can VRS have a negative impact on your documents, it can also become an issue for the physician. In order to use voice recognition software, physicians must speak louder and more intensely than they usually do. As a result, physicians may experience some serious voice strain. Voice strain induces hoarseness and, in worst case scenarios, can lead to vocal hemorrhage or a stroke. This happens in the same manner that excessive coughing, blowing one’s nose too hard or sneezing occurs.
Another issue with VRS is it’s sensitivity to sounds (other than the physician’s voice) in the room. Those physicians who see the best results with VRS typically use a background-noise-filtering microphone. Then again, even in the perfect atmosphere, this technology isn’t yet able to perfectly distinguish the speaker’s voice from other voices in the room.
At best, these noises can be a distraction that may impact your ability to concentrate on what you are doing. Not to mention the confusion your staff may experience when looking into your electronic health records (EHR) and finding random words or sayings picked up by your VRS.
HIPAA compliance issues
When all is said and done, proper medical documentation isn’t just about avoiding errors. It’s also about remaining compliant with the proper rules and regulations. With the coming of age of EHR, it’s becoming harder and harder for practices to remain HIPAA compliant. HIPAA regulations remain the standard for patient information protection. EHRs alone make compliance difficult for many practices. Add in the potential errors and data issues that come with VRS and compliance is even more difficult to achieve.
So how can your practice remain HIPAA compliant and achieve fast accurate transcription? The answer is DataMatrix Medical. Not only do we offer voice recognition proofreading, but also medical transcription services that can transform your practice and its functionality.
We provide a team of human medical transcriptionists to each of your physicians to enter data spoken by the physician into your EHR–all while maintaining your patient data security.
At DataMatrix Medical, we provide fast, accurate solutions to physicians who want to spend less time entering data into a computer, and more time with their patients. If you’re ready to learn more about our solutions, download our free case study below. In our case study, we detail the daily life of one of our clients before, during and after starting our free, two-week medical transcription trial. Click the button below to download your copy, or contact us for more information!
Content was originally published on January 20, 2018. Content was refreshed on July 9, 2019.