Speech recognition and direct data entry in clinical microbiology

Area of Science:

• Medical Laboratory Science
• Health Informatics

Background:

• Traditional data entry for laboratory results is time-consuming and prone to errors.
• Existing methods often involve manual coding and keyboard input, delaying report turnaround times.

Purpose of the Study:

• To design and evaluate a speech-based system (Vigilante) for direct data entry of urine microscopy, culture, and sensitivity test results at the bench level.
• To reduce data entry time, improve accuracy, and expedite report issuance.

Main Methods:

• The Vigilante system was developed for natural continuous speech input, incorporating graphical and audio feedback for error correction.
• A three-month comparative study involving 2360 urine specimens evaluated Vigilante against traditional data entry methods with multiple operators.

Main Results:

• Vigilante significantly reduced data entry time compared to traditional methods.
• The system demonstrated improved accuracy in reporting and decreased overall report issuance time.
• Graphical and audio feedback features facilitated efficient error recognition and correction.

Conclusions:

• Direct data entry using speech technology, as exemplified by the Vigilante system, provides a practical, reliable, and innovative solution for laboratory reporting.
• The Vigilante system is now routinely used for urine test results and is adaptable to other laboratory methodologies and data entry tasks.