Speech Recognition – It’s Time To Move On? Leave a comment

This article was automatically transcribed using Dragon speech recognition.

Today’s NHS trusts are all being put under massive pressure to improve efficiency and cut costs. The directors and managers within every hospital have to consider and look for ways of reaching their targets. It’s a very tough balance to spend money on improving services and cutting budgets to save money. One thing that all hospitals have in common is their production of documents and correspondence from the doctor to the patient. For several decades this process has been handled by a doctor dictating a letter onto a tape and then passing this to the secretary. Digital dictation and speech recognition are two solutions which can dramatically improve the efficiency of this process. The benefits being that once a patient has been to a hospital and had an assessment, their GP and themselves will be notified sooner and ultimately patient waiting times are reduced.

Generally speaking it is now very common to find speech recognition used within radiology departments in hospitals across the country and for isolated departments to have moved from analogue dictation to digital dictation. It is also common to find departments within hospitals using outdated analogue equipment which is causing a major delay to their correspondence. A lot of case studies will show that a typical delay for a trust with analogue dictation in place can be anywhere between 4 to 8 weeks. This can be down to a number of considerations including the fact that the tape has to be physically moved between the doctor and where the secretary is based. The quality of the sound of the tape is also much lower than digital and it only gets worse after every recording and deletion. Tapes can also be lost, damaged or be inaudible for the secretaries. Resulting in a doctor needing to dictate the letter again. Not easy when the original letter was dictated several weeks before!

The introduction of digital dictation in hospitals who have an eight-week delay between when the doctor dictated a letter and the letter being ready to be sent to the patient and GP can be quickly reduced to one or two days. This seems too good to be true, but this is the real truth of how dramatic the improvement can be. The doctors do not talk any faster and the secretary does not type any faster, so how can a letter be typed so much faster? Dictations are sent instantly from when the doctor finishes each job, and is then available in a clearly laid out work list on the secretaries PC. The doctors are able to clearly mark priority dictations which the secretaries can then type head of standard dictations. Many digital dictation solutions will also encrypt the sound file to ensure that patient confidentiality is protected. This is impossible to do with a tape.

Speech recognition can be used to work in parallel with digital dictation or as a total replacement. As I’ve already mentioned, most radiology departments within hospitals have been using speech recognition since it became available sometime ago. The use of speech recognition suits them perfectly as they are able to review images and results and into their analysis into the report using their voice. Speech recognition allows them to do this in real time and complete and sign off reports instantly. Other departments around these hospitals can also benefit from using speech recognition in order to complete forms, letters and patient reviews. Ideally, the user would have access to a PC or Windows-based tablet to allow them to use front-end recognition directly into their system. However this can also be done in the backend.

The speech recognition market has had constant and healthy growth within the healthcare sector for the last 15 to 20 years. With the advance in technology, improved spec of PCs in hospitals and research into speech recognition, the current solution is now better than ever. Many people who may have tried speech recognition in the early days may have had a bad experience as the software did not meet their expectations. The latest versions of the software are now miles away from the accuracy that was achievable even two or three years ago. The developers have worked very hard to ensure that the product works to a high standard straight out of the box. Of course, training will improve the accuracy and it will never be 100% error-free, but it won’t be far away. It is always recommended that the users of the software should be fully trained and their speech recognition profile customised in order to get the most from it. It would be a mistake for any trust to invest in speech recognition without having also invested in training and support from a specialist supplier. I have seen that many project managers are tempted to cut back on this relatively small expense in order to save money but then they do not get the full benefits of the solution they have just paid for. It’s like buying a fast car and not taking it out of second gear!

The question that many trusts should now be asking themselves is can they afford not to have digital dictation and speech recognition in use across their whole trust. Continuing to use outdated analogue dictation is costing them money. It’s now time to move on.