Ohio hospital shows advantages of speech recognition in medical reporting

Mount Carmel St. Ann’s Hospital uses ScanSoft Dragon NaturallySpeaking Medical

Mount Carmel St. Ann’s Hospital is a large healthcare provider near Columbus, Ohio, and part of Trinity Health, the third largest Catholic healthcare system in the United States. The hospital has implemented the ScanSoft Dragon NaturallySpeaking Medical 8 speech recognition solution to automate transcription within its Emergency Department. Using speech recognition instead of manual transcription of recorded medical reports, the emergency department has reduced transcription services by 97% and is saving $600,000 per year. Part of the reasons for the large savings is that doctors are doing their own editing, largely on Tablet PCs, immediately after the report is generated, according to a spokesperson at Mount Carmel, as opposed to having the unedited speech-recognition-generated report reviewed by a transcriptionist later. The speech recognition software is currently on each PC, with voice files saved on portable media; but the hospital is examining the possibility of a networked system.

Patient care is improved because the medical record is available immediately rather than typically 24-48 hours later when manual transcription is required. Doctors can review the report immediately after seeing the patient and without many more intervening cases; logic would suggest that a report reviewed immediately after dictation, when the case is fresh in the doctor’s mind, will be more accurate than one reviewed over a day later. Dr. Loren Leidheiser, chairman and director of emergency medicine, confirms this point: “Using Dragon NaturallySpeaking has… greatly reduced the number of errors in patient records. All of this translates to improved care for our number one priority—our patients.”

Dragon NaturallySpeaking Medical supports dictation of patient notes, reports, letters, and other documents. The healthcare professional can dictate directly into a microphone attached to a PC or to a tablet computer, or into a portable digital voice recorder and download the speech onto a PC for transcription. Mount Carmel chose to make the system available on rolling carts mounted with wireless PCs and portable handheld tablets.

NaturallySpeaking Medical has a built-in statistical language model reflecting medical vocabularies and phrasing. Macros supported by the software can accelerate reporting by allowing several operations to be completed by one voice command or an entire standard paragraph to be inserted with one spoken phrase. Templates allow the doctor to fill in a form rather than worry about the formatting of the report (and also serve to remind the doctor of information that should be included in the report). Leidheiser estimated getting recognition accuracy of more than 98%. The more complete recording supports more complete billing, with the hospital estimating an average of $20 additional billing per patient visit, a total of $1.2 million in the first year.

Reports can be automatically inserted into the hospital’s Electronic Medical Records (EMR) system. Mount Carmel uses the A4 Healthmatics ED EMR application.

NaturallySpeaking Medical 8 has a built-in vocabulary of more than 300,000 words, and includes 14 pre-made specialty vocabularies, including Emergency Medicine, General Medicine, Pathology, Radiology, Cardiology, and Surgery disciplines. Healthcare organizations can also add their own words to create fully customized vocabularies.

John Shagoury, president, ScanSoft Productivity Applications Division, said, “Whether in an individual department, such as Mount Carmel St. Ann’s Emergency, a small practice, or across an entire healthcare organization, ScanSoft Dragon NaturallySpeaking is proven to dramatically reduce costs associated with transcription services and other repetitive tasks, boosting productivity and freeing up healthcare professionals to focus on patient care.”