Saturday, December 15, 2007

Voice Recognition and Value

Dr. Giles Boland has written an editorial about the value radiologists can provide with speech recognition (Dr. Boland prefers the more popular though less accurate term Voice Recognition). Dr. Boland is on the radiology staff of Mass General, and he is a member of RCG HealthCare Consulting, which provides "Complete IT & Management Consulting for Radiology Departments and Imaging Practices". From the RCG website, we find that Dr. Boland is:
Vice Chairman of Radiology, Massachusetts General Hospital, Dr. Boland’s areas of specialty include PACS, Teleradiology, Voice Recognition, RIS, and the enterprise digital solution to PACS and RIS integration. He has conducted aproximately 40 pesentations in over 15 countries on these topics. As a practicing radiologist, his nterests lie in Abdominal Imaging and Interventional Radiology. Dr. Boland is an Advisor to the World Health Organization, Geneva Switzerland and a reviewer for multiple scientific journals including New England Journal of Medicine, Radiographics, American Journal of Roentgenology and Journal of Intensive Care Medicine.
Those are impressive credentials. Dr. Boland's case for VRT is as follows:

Voice recognition technology cuts a swath across the process through which conventional preliminary findings metamorphose into a final report. Once a report is dictated into VRT, it is in fact a final, signed report (unless originally dictated by a radiology resident or fellow). By virtue of its electronic nature, such a report becomes available immediately across an institutional network, simultaneously to multiple caregivers. Consequently, final report turnaround times are typically drastically reduced. When VRT was introduced at the Department of Radiology at Massachusetts General Hospital in 1997, the final report turnaround time for staff dictated reports was reduced from 3 days to several hours almost immediately [11]. This efficiency was realized despite the fact that earlier VRT models were harder to use and had less efficient speech recognition software.

The implication is that the technology has improved a great deal; the closing paragraph in the editorial reads:

However, despite the real advantages to radiology customers of VRT, some radiologists would still rather promote an inferior transcription model, preferring instead to use traditional dictation methods, which delay their ability to generate final reports. Although radiologists' customers are looking for succinct, standardized, and timely final reports, some radiologists continue to use a system that their customers find less valuable. Rather than using existing state-of-the-art technology, radiologists should take an active role in convincing their peers to adopt VRT. If necessary, they should also lobby their organizations to provide the capital required to finance the transition, which generally yields a very favorable return on investment within the first year [11]. Radiologists can then rightly claim that they have been instrumental in adding significant value to their product, a major benefit to patient care and all stakeholders.
Note the derrogatory language. Those that have not embraced VRT/SR are using an inferior model. This raises some concerns. Whilst Dr. Boland has the background to know what he is talking about, the superior attitude is not particularly endearing. We have had the VR/SR debate on AuntMinnie ad nauseum, and the majority opinion amonst radiologists (NOT administrators, IT types, etc) is that it is not ready for primetime. Maybe Dr. Boland has access to more advanced software that actually works as advertised, which would make his analysis spot on. Unfortunately, the machinery that makes it out to the boonies doesn't seem to work well enough to justify the accolades. I think it is noteworthy that the JACR article contains no mention of Dr. Boland's association with RCG Consulting or the fact that RCG Consulting considers voice technology one of its areas of expertise. Perhaps it is simply understood that anyone on the Mass General Radiology staff is a part of RGC. I'm sure there is no conflict of interest here. Of course not.

I'm not so much of a Luddite that I don't appreciate what Dr. Boland is saying here. I just don't think the machinery is quite there yet. So, I'll stick with my inferior model for now, thank you.


Anonymous said...

Right on. That editorial struck me as condescending, too.

So many of the articles singing the praises of VR come from academic or hospital-based radiologists. Some apparently have ulterior motives for promoting VR, being consultants or paid lecturers on the subject. Other radiologists promoting VR have had VR thrust upon them by their hospital administrators, who see only the benefit of firing all transcriptionists without the cost in radiologists' time (which after all is free to those hospital administrators). Some radiologists in this latter category figure, "When life hands you lemons, make lemonade. Maybe we can at least get some papers and presentations out of this bad situation." Finally, some radiologists switch to VR from a transcription environment that is truly atrocious -- inaccurate reports requiring many corrections, several-day turnaround times, or a high turnover rate in the pool of transcriptionists such that few get accustomed to the dictation habits of individual radiologists.

Our practice fits into none of those categories. We are a freestanding radiology group. We employ our own transcriptionists, most of whom have been with us over a decade. We have quick, accurate transcriptions, and I frequently see reports that I have marked "STAT" appear in the queue within five minutes.

The real question for me is, "Is there a group out there in our same situation that has switched from traditional transcription to VR and is happy they did so?" I have yet to encounter such a group. Most of the demonstration sites promoted by the VR makers are hospital-based -- someone else was paying the transcriptionists, and someone else made the decision to switch. Any advice those sites can give me is of limited utility.

Yes, the transcriptionists' salaries are ongoing and considerable, and they come out of my pocket. Yes, it'd be nice to reduce that. And yes, I understand that while radiologists' time is valuable, it's not infinitely valuable. Nevertheless, I remain unconvinced that doing away with transcriptionists altogether makes economic sense, particularly since all of us are only going to get busier in the next several years. I'd be willing to explore the possibility that VR gets the first crack at the transcription, changing our transcriptionists into editors, but for me it's gotta be a black box.

Anonymous said...

Hi Sam,

Dig deeper. Individuals at MGH profited greatly by embracing and facilitating the start and use of Talk Technologies (sold to AGFA) and Commissure (sold to Nuance). The same people at MGH and Mike Mardini were involved with both.

I'm not implying anything wrong, there isn't/wasn't. I'm just stating that it sure is nice to have "Man's Greatest Hospital" get behind technology, sell it, and profit from it.

Anonymous said...

The writing on the all is there: Hospitals will ram VR down all our throats thanks to these academic types (who probably have residents pulling their hair out trying to get VR to work).

Anonymous said...

Nice blog. Well said about residents pulling their hair out in the comment above. We actually nicknamed our software "TypeTech." By the way, I started a "Voice Recognition to English Dictionary". Please feel free to contribute your translations...