Speech recognition and direct data entry in clinical microbiology

Abstract

The aim of this study was to design a system (Vigilante) for the direct entry of urine microscopy, culture and sensitivity test results at bench level using natural continuous speech, removing the need for time-consuming report coding and keyboard entry. Graphical and audio feedback facilitated error recognition and correction, and removed the need for time-consuming report checking and validation stages. The system was evaluated over a three-month period by direct comparison of 2360 urine specimens processed by multiple operators using both Vigilante and traditional methods. The results showed that the Vigilante system reduced the time taken to enter data, improved the accuracy of reporting and reduced the time taken to issue reports. The results suggest that direct data entry using speech offers a practical, reliable and innovative alternative to more conventional means of data entry. The system is now used routinely in this laboratory to report urine microscopy, culture and sensitivity test results, and may be adapted to individual urine methodologies or other data entry tasks.