KLAS rates back-end, front-end speech recognition vendors

Workflow efficiencies and cost savings benefits are speeding speech recognition technology adoption.

Prompted by the movement toward speech recognition as the next step in automating data flow and documentation, KLAS has issued its Speech Recognition report, which analyzes both back-end and front-end speech recognition technology vendors.

For the report, KLAS interviewed more than 300 healthcare professionals regarding the speech recognition systems in use at their respective organizations.

For front-end speech recognition, Nuance received top ranking for its RadWhere product in overall performance scores, followed in performance ranking by Nuance (PowerScribe), MedQuist (SpeechQ) and Agfa (TalkStation).

eScription EditScript (now Nuance) received top ranking in the back-end speech recognition category for overall performance scores, followed in ranking order, by: Dolbey (Fusion Speech) and Nuance/Dictaphone (EXSpeech).

The main vendors and products included in the study were validated at no fewer than 15 unique inpatient organizations. According to KLAS, the data collected for the study reflects hospital centric experiences with these systems.

The majority of survey participants included: radiology/imaging director/managers (22 percent); IT director/manager (20 percent); HIM director/manager (16 percent); and transcription director/manager (15 percent).

KLAS reported that all vendors in the study utilize a common set of speech engines, including Dragon from Nuance and SpeechMagic from Philips. The differences lie in the bolt-on technologies built on top of the speech engine, such as various user interfaces, administrative tools and provider-driven workflows, according to the report.

“Beyond these features and functionalities, differing client satisfaction stems from the way the vendors interact with the client via the sales force, support infrastructure and executive team,” the report noted.

According to KLAS, of the three back-end speech recognition vendors, eScription is the “clear leader in client satisfaction,” due to solid service and knowledgeable staff personnel. eScription received 91.4 out of 100 possible scoring.

For front-end speech recognition, Nuance’s RadWhere received 87.2 out of 100 for its performance score.

Additionally, the report revealed that the performance score of 82.2 out of 100 for the back-end speech recognition vendors was higher than the average performance score for the front-end speech recognition vendors, at 78.8.

KLAS stated that the reason for higher back-end scores is due to easier adoption and minimal impact on physicians. Respondents stated they introduced back-end instead of front-end technology to minimize interruptions in physician workflow, despite an enrollment process that must still take place when going live with the technology.

In contrast, physician adoption and satisfaction are chief concerns of survey respondents regarding front-end speech recognition technologies, according to the report.

Additionally, survey participants reported that the top three benefits to front-end systems were  improved turn-around times (76 percent); decreased costs (28 percent); and staff reduction (17 percent).

For back-end technology, the top three benefits were productivity (69 percent); improved turn-around time (61 percent); and decreased costs (49 percent).

According to the report, although radiology still represents the major field of use for front-end speech recognition systems, hospital managers intend to expand the utilization of these systems outside of radiology. Among the first department managers targeted for expansion is cardiology, although KLAS said it has validated “only a small number of organizations” that are using front-end systems in cardiology.

The next three to five years will be interesting for the speech recognition market, according to the report, as radiology departments debate whether to choose a standalone speech recognition vendor or to purchase a PACS with embedded speech recognition.

“The burden will be on the PACS vendors to prove their speech recognition and documentation systems are functionally adequate, user friendly and that the integration feature is a true benefit,” KLAS concluded. “The burden for dedicated speech recognition vendors will be to show they can truly facilitate departmental workflow to a greater degree.”

More information about the report is available here and through KLAS’ website.