Twitter Updates

    follow me on Twitter

    16 September 2006

    Digital Dictation and Voice Recognition: Not ready for prime time?

    Our digital dictation system consists of a Philips Speech Mike linked in to our RIS (TC-RAD, McKesson). For plain film reporting, after using it for a year or two, I have switched back to using cassette tapes, and guess what? I estimate that my productivity is up 20%, and fatigue levels reduced by 50%. Other colleagues have reported similar experiences.

    The problem with our digital dictation system is that it involves numerous mouse clicks and aiming a cursor several times during the dictation process, and this increases the total amount of work very substantially if the report is very brief (e.g. "Normal appearances"). Also, transcription of reports by secretaries from digital dictation involves more steps and is inherently less efficient than transcription from tape.

    There are advantages to digital dictation, and it's OK for longer reports like CT and MRI reports. It allows secretaries to retrieve and locate dictated reports more easily. However, at least in our version, there are very serious inadequacies.

    As for voice recognition (VR), I tried that too, for several months, and put enormous effort into training the system, which was created by bolting the Dragon dictation system (which seems to have nothing but outstanding reviews in PC magazines) onto our digital dictation system. I came to the conclusion that I would have been better served by doing a touch typing course.

    Here's an online article that describes in great detail the experiences of many users with radiological voice recognition systems. I concur with the opinions expressed.
    Voice, I mean Speech Recognition, Buggy Whips, and OODA Loops

    If VR makes sense on an accountant's spreadsheet in terms of staffing levels and turnaround times, this may be because the spreadsheet has not recorded all of the variables. I am now extremely sceptical about the practicalities and costs of these systems in their current form.

    From personal experience I am wary of verbal or published recommendations for VR systems and other software products, especially those designed for medical use, where standards and expectations are often extremely low. Someone who has spent vast sums of money on a poor quality product may be reluctant to admit (even to himself, let alone to the outside world) that it was all a big mistake.


    Dalai said...

    Thanks for linking to my article. Yours adds to a growing number of negative reports about VR. It just isn't ready for primetime, is is?

    Dominic Blunt said...

    If VR/SR is not working for you, it is simply because you have the wrong product, badly implemented. Simple as that. In our 1000 bedded teaching hospitalin London, with about 40 consultant staff, and 31 trainees we implemented it as a 'no-option' big bang nearly 2 years ago. Our productivity rocketed, this has been sustained, and its accuracy is superb for most of us. we will be publishing our data and details shortly. Delighted to show anyone who wants to come and see!

    Anonymous said...

    have you ever tried digital dictation on a mobile device instead of a desktop microphone?
    As far as I'm concerned, that makes things quite a bit easier (saves all those extra mouse clicks...;-))
    And our transcriptionists are still using their analog equipment for typing, so I really don't see any problem (for digital dictation that is, speech recognition is a chapter of its own...)

    Here's my personal current fave on mobile devices: