ORLANDO, Fla., Feb. 23 /PRNewswire/ -- At the 2004 HIMSS Conference, ZyDoc announced evidence confirming the impending widespread adoption of speech recognition in the medical industry incorporating ZyDoc specialty specific vocabularies. Multiple vendors are now delivering speech recognition applications offering highly accurate ZyDoc continuous speech recognition modules for medical documentation.
James M. Maisel, M.D., Chairman of ZyDoc described the history of speech recognition in the medical industry, "In 1994, ZyDoc released one of the first Windows based multimedia medical record systems for beta testing. Although fulfilling most of the recommendations of the Institute for Medicine for CPRs, we identified the data entry bottleneck that the industry still faces today. ZyDoc selected speech recognition as the technology that would allow doctors to document medical encounters quickly with little change in work habits and began developing language modeling initially with Philips Speech Processing and then independently."
Although ZyDoc released speech recognition language models, contexts and vocabularies with breakthrough accuracy in multiple medical specialties in 1998, adoption in the industry was slow. ZyDoc was recognized with awards from the computer industry in 1999 and endorsed and bundled by Dragon Systems with their NaturallySpeaking Medical in 2000. Subsequently, the industry began to recognize that highly accurate speech recognition required specialized language models and vocabularies in addition to the generic speech recognition engines. However, this was only a component of the systems needed by physicians. Speech recognition required applications to allow management of the text output.
The tumultuous shakeout in the speech recognition industry impeded development of speech driven medical applications until 2001 when ZyDoc licensed its Speech Medical Vocabularies to Talk Technology and other vendors. The Talk Technology application, TalkStation, achieved highly accurate speech recognition using the ZyDoc language vocabularies. It received immediate physician acceptance with installation of over 2500 workstations in Radiology alone and was recognized by Klaus Enterprises as the leading speech recognition vendor. ZyDoc now offers its own third generation, multimodal speech application for report generation, SpeechDoc EMR 3.0.
Widespread dissemination of speech recognition applications is now taking place in the industry largely due to the highly accurate ZyDoc language models or vocabularies inside of speech recognition applications. Significantly, Talk Technology, a subsidiary of Agfa-Gevaert Group, has now extended the breadth of speech recognition applications incorporating the ZyDoc speech vocabularies in cardiology and emergency medicine in its TalkStation Cardiology and EM (emergency medicine) applications. To accelerate implementation, Agfa enlisted several healthcare IT vendors as partners, including Cerner Corp, IDX Systems Corp, GE Medical Systems, Siemens Medical Solutions and MedQuist.
According to Dr. Maisel, "The real acceptance of the technology is now marked by the adoption of speech recognition by the transcription industry using the ZyDoc software. Our experience with hundreds of physicians in multiple specialties identified many issues other than accuracy and applications that were impeding adoption of the technology. To further accelerate industry usage, ZyDoc has subsequently released vocabulary upgrades, and offers its own turnkey speech recognition SpeechDoc application bundled on Dell and Toshiba workstations. ZyDoc also has an integration program helping EMR companies embed the speech recognition technology, Web- based training, a strong reseller channel and low technology solutions for non-technical doctors."
"We are pleased to offer our most significant upgrades to current users of the Version 5 language models currently used in the industry," Dr. Maisel announced. These medical specialty speech vocabularies and topics are now available for Dragon NaturallySpeaking Professional 7.0 and IBM ViaVoice 10. The upgrades are quite significant due to the updated terminology based upon enormous amounts of real life data available from the transcription arm of our company. The technology offers such accurate recognition that dictation errors usually exceed any speech recognition errors. The bigram and trigram models incorporate AAMT preferred style of terminology expression, eliminate most contextual errors and include the latest medications, equipment, procedures and medical terminology by specialty. Even modest improvements for physicians from 97% to 98% accuracy mean 33% fewer corrections for physicians and large productivity gains."
