Augnito/ Scribetech

Just to add my 2 pence – I’ve just switched on Augnito in InSight PACS based reporting. Hoping to roll it out to reporting rads next week and Radiologists will follow when they’re happy.

So far, seems awesome. Really fast and great recognition, plus possible to control all PACS too/ls/actions with it.

I’ll feed back when my clinical staff have used it in anger, but from a PACS point of view I’m very impressed.

I’d agree with James Ricketts comments. Used it at Circle/ BMI with CRIS.

I think it is better than Dragon but the faults he outlines are irritating and one would have thought not too difficult to solve.

Little update. So far so good. All our Reporting Radiographers are using it now, plus 2 early adopter Consultant Radiologists (one’s a very keen VR user and the other was having such major problems with DMPE we decided to switch him early).

Recognition seems very good out of the box, across a variety of accents, with a few exceptions that users are getting used to (eg. one user said “Get red star” and Augnito typed “Get red stop” - we’ve built a new coded phrase for “Get red stop” that types exactly what they want to appear). They would like the ability to train some words, but that’s coming in the next PACS release in a couple of months.

The responsiveness is excellent. No issues with speed of recognition across a variety of PCs using the hospital internet connection, even for home users where the connection goes via a VPN back to the hospital first, then out to the internet. There’s a little indicator in PACS to show when Augnito is listening to you speak, then the second you pause for breath all the text appears.

We’ve set up plenty of canned text phrases, linked to voice commands, Both those pushed out to all users and those set up individually seem to work fine and follow users around the various workstations they use.

Have to say from an administrative point of view I’m very pleased with Augnito and its integration into InSight PACS. Setting a user up is literally a single tickbox in the user settings. I’ve not had to deal with any of the Dragon-esque issues, like corrupted or bloated profiles, profile roaming not working, etc. So far, Augnito has just worked.

We’ve also started testing the use of single profiles across multiple Trusts and PACS systems, where a user may work for multiple institutions. It works for me and we’ll try with some clinical users soon.

We’re going live with the rest of the Radiologists next week - I’ll provide another update once they’ve had a few weeks to get their teeth into it.

like WANG, Alan (GREAT WESTERN HOSP… reacted to your message:

Another update a few months down the line.

I would call it a promising start, but there’s plenty for Insignia to work on with Scribetech to make the reporting experience more streamlined.

The recognition seems to be, on the whole, very good across the vast majority of users. A couple of users were really struggling, but having replaced their speechmikes with Logitech Blue Yeti microphones they are now happy. I’m wondering if the integrated positional audio of the Yeti mic has something to do with the better recognition, but equally it could be that the speechmikes had become gunged up over the years…

There are several complaints about grammar and navigation. Currently, you can jump between template headings ("Go to Clinical History/Findings/Impressions/Summary) but we aren’t yet able to move between fields in a template. Also, there seem to be some random capitalisations, or lack thereof after a colon, and if you pause before saying “full stop” at the end of a sentence, it leaves a space. The bit that the seasoned Dragon users are missing though is the ability to train their profiles.
I’m told these should all be fixed in the June release of PACS, including training profiles and adding words to the vocabulary - I will feed back more then.

As mentioned before, the admin side is worlds apart compared with Dragon. The voice recognition just works, there are no corrupt profiles or loss of commands to deal with, no management of profiles across various locations, no need to run optimisation, etc. Other than showing our reporters how to turn on VR and add their commands, there has been no admin effort required at all.

I’ll update again when we’ve got the June update in.

All the best

Andy

2 Likes

Has anyone used a VR product such as Dragon Medical One sitting outside of the RIS and essentially pasting the reports into CRIS rather than relying on the integrated version of Augnito/Dragon and using it to drive the whole PC as opposed to just CRIS? This is a path we are currently exploring due to version issues between Red Hat/Windows/CRIS/Dragon.

Yes, we do this at SASH currently.

With DRAGON stand alone, because we don’t have integrated VR in RADNET.

Personally, I think that Copying and Pasting clinical information between separate systems could present you with an IG risk e.g what are the implications if all of the text isn’t highlighted and copied across correctly?

DMO does this automatically with the command “transfer text”

The risk that happens far more often is you end up loosing the text you intended to transfer but there is an undo button. After a bit of getting used to, this isn’t a big problem.

The voice recognition engine is excellent and based on my testing far better usability than some more integrated solutions (taking into account the lack of integration).

Cost might be a barrier.

1 Like

For various reasons we’ve ended up with users implementing many different workflows within our department.

  1. ‘Approved’ Dragon VR into a 3rd party product which integrates with CRIS
  2. Dragon dictation pad, then manual cut&paste into CRIS
  3. Trust’s M*Modal Fluency Direct dictation pad, then manual paste into CRIS
  4. MS Office dictation into Word, then manual cut&paste into CRIS
  5. Typing into CRIS

Obviously support is a bit of an issue, but the biggest problem is we’re now unable to define/mandate the use of report templates to support standardised reporting, Failsafe Alert triggers, etc.

Most VR products are able to bind to a specific application window or control so that they can work in the background and the VR text just ‘magically’ appears in the target app. Unfortunately, I’ve not managed to find a way to get this to work with the Java-based CRIS Core…

We’re therefore waiting for the first one of these:

  1. CRIS Core/Augnito integration with acceptable functionality
  2. CRIS Evo (Reporting and Vetting modules) integrated with Fluency Direct
  3. PACS-based reporting
  4. I retire

Should run a sweepstake really :slight_smile:

2 Likes

I agree with Stephen’s Gallagher’s comment. It appears to me that you, like us, are searching for ways of getting the CRIS product to work reliably. You are trying to solve, in my view, what is your suppliers responsibility. I’ve tried for 3 years, many upgrades, deep dives (still ongoing), knee deep frustration and various radiologist suppressed mutiny attempts later, I am pushing for option 3 or 4 :grin: on your list (whichever comes first).

2 Likes