Preliminary Evaluation of Speech Recognition for Capturing Patient Information at Nursing Shift Changes: Accuracy in Speech to Text and User Preferences for Recorders

Select |




Print


Suominen, Hanna; Basilakis, Jim; Johnson, Maree; Sanchez, Paula; Dawson, Linda; Hanlen, Leif; Kelly, Barbara


2013-02-11


Conference Material


The 4th International Louhi Workshop on Health Document Text Mining and Information Analysis (Louhi 2013)


Sydney, Australia


-


Verbal communication at nursing shift changes provides an accurate representation of patients’ background and current state of clinical management. However, over two thirds of this valuable handover information is lost after three to five shifts if written notes are taken by hand or not taken at all. These failures in communication cause the largest number of sentinel events in the USA and nearly half of all adverse events and over a tenth of preventable adverse events in Australia. This paper studies the use of speech recognition to automatically transcribe verbal handover information into written text by evaluating accuracy in speech to text. We also evaluate user preferences for reordering devices. These evaluations in simulated clinical settings compare six recorders and five microphones on ten scenarios from nursing shift changes. For the accuracy evaluation, we use professional-level devices and studio for recording three people and playing the recordings across all person-recorder-microphone combinations. We study five clinicians’ preferences via an eighteen-item pre-survey, eleven-item post-survey, and one-to-one interview. The pre-survey ad-dresses initial perceptions of using speech recognition in clinical settings: un-derstanding of the related technologies and perceived benefits/problems with the clinical application. The post-survey focuses on perceived usability of the same recorders and microphones as in the accuracy evaluation. Digital voice recorders result in the best accuracy (almost 79 per cent correct) when compared to the original scenario by using off-the-shelf software with minimal tailoring to a speaker. Accessory microphones improve the recorders’ accuracy. Noise can-celling lapel microphones give promising results and also four out of the five clinicians prefer lapel microphones over headsets. The clinicians’ most prefer recorder is a smart phone followed by a tablet computer, digital voice recorder, and MP3 player. These results encourage cascading speech recognition with written natural language processing in order to capture verbal handovers as document drafts for subsequent clinical review, editing, and addition to com-puterised medical records systems.


Continuity of Patient Care; Documentation; Evaluation; Medical Records Systems, Computerized; Speech Recognition Software


http://nicta.com.au/business/health/events/louhi2013


nicta:6423


Suominen, Hanna; Basilakis, Jim; Johnson, Maree; Sanchez, Paula; Dawson, Linda; Hanlen, Leif; Kelly, Barbara. Preliminary Evaluation of Speech Recognition for Capturing Patient Information at Nursing Shift Changes: Accuracy in Speech to Text and User Preferences for Recorders. In: Hanna Suominen Editor, editor/s. The 4th International Louhi Workshop on Health Document Text Mining and Information Analysis (Louhi 2013); Sydney, Australia. 2013-02-11. -. http://hdl.handle.net/102.100.100/98173?index=1



Loading citation data...

Citation counts
(Requires subscription to view)