http://www.ihealthbeat.org/features/2011/nine-experts-weigh-in-on-2011-health-it-progress-hopes-for-2012.aspx

http://www.healthcareitnews.com/news/5-tips-speech-recognition-software-rsna

However it did produce significant results in certain areas. Radiology transcription is a case in point. Recent survey conducted by Digital Imaging shows that 80% of the respondents use voice recognition technology but 30% expressed dissatisfaction with the results.  Radiology reports, more often than not, are short and the same content is repeated across documents. Radiologists can train themselves in using the speech recognition technology in the front end and also create normals and ‘call’ these normals while dictating. This makes the job significantly easier and does not take too much of the radiologists’ time . Hence you see that the adoption rates for speech recognition technology in radiology space is much higher than in other places. 

Use of speech recognition technology in medical records is a completely different ball game. Here the success rates vary widely across hospitals, providers and work types.  Some physicians  in hospitals are in love with it while others have really given up hope.  When using back-end speech recognition, providers who make an effort to dictate clearly, speak complete sentences without mumbling or and try not to dictate at rocket speeds, often experience good results. Not surprisingly they are also the easy ones to manually transcribe. Physicians with ESL, or those that have bad dictation habits experience poor results. They are also the most difficult to transcribe manually. Different work types also give different results. Highly structured and templated documents often provide better results than complex reports using long free flowing text.

I think the solution to this debate of whether this technology has ‘arrived’ or not, lies with the technology providers in setting the expectations correctly. They should not over sell the technology and create a hype which gives the impression to the HIM and provider staff that this is the magic bullet that everyone has been waiting for. Clearly this technology has limitations and customers need to be educated about them. Currently customers expect the solution to provide excellent results out-of-the-box. But results are directly proportional to the effort one puts in, in terms of committment to training themselves, being patient with the system while it is ’learning’ the speech pattterns and algorithms, and are ready and willing to change some of their dictating habits.

Speech recognition technology has come a long way but it is still not at a point where we can expect it to give excellent results consistently.

In an era where clinical documentation is generated and maintained in an ‘episode’ nature, EHRs pose significant challenges especially when you have to share the data. EHRs maintain data in a ‘continuous’ care mode and therefore it becomes challenging to decide from which data point should you share the information to other stake holders in the care domain. Challenges also emanate from the gradual shift towards ‘charge for care’ rather than ‘charge for service’ model.

Contrary to conventional thinking, the advent of EHRs has actually brought the entire clinical documentation area into the front and center of patient care.