Abstract
Background. Speech recognition (SR) technology has been successfully used in many specialties, particularly in radiology. However, there have been only a few studies validating its feasibility and favorable system configuration in cardiovascular medicine. Ambulatory electrocardiography (AECG) is one of the most frequently ordered examinations in clinical practice, and its results are ordinarily returned in the form of analysis report by cardiologists. To determine more favorable SR system for AECG reporting generation in terms of time, we evaluated two SR systems using different types of dictating microphones.
Methods. The SR software used was a commercially available SR system called AmiVoice Ex7. One hundred forty AECG reports generated via a conventional USB handmike and 153 reports using a Bluetooth wireless headset with an enhanced reference lexicon associated with cardiovascular medicine were assessed. Report turnaround time includes additional time for correction of recognition errors.
Results. Turnaround time by the headset system was statistically shorter than that by the handmike system (median : 615.5 sec vs. 516.0 sec, p=0.001). Furthermore, this tendency still largely existed regardless of the number of findings associated with significant arrhythmia and myocardial ischemia. Conclusion. Considering the hands-busy setting of AECG analysis, the hands-free SR system specialized for cardiovascular medicine might be more time-efficient and comfortable than exsisting handmike systems in generating electronic AECG reports.
