JMIR Publications

ABSTRACT

Background:

Nursing records are crucial for maintaining patient care quality but impose a significant workload on nurses, diverting attention from direct care. Speech input technology offers hands-free documentation, enabling simultaneous patient care and record-keeping. Despite its advantages, adoption has been slow due to concerns about patient privacy, technical challenges, and integration complexity with electronic medical records. Variability in nursing workflows and system adaptability further hinder implementation. These challenges have resulted in few successful cases being reported. This study examines the application of speech input technology in angiography rooms, which offers controlled environments conducive to testing due to their highly standardized workflows, patient privacy, and screen-free interaction requirements.

Objective:

This study is aimed to assess the feasibility of speech input technology in real-world clinical settings, focusing on its usability, effectiveness, and adaptability. Additionally, we seek to identify barriers to broader adoption and propose actionable strategies to overcome these challenges, enabling its integration into routine nursing practices. By addressing these barriers, we aim to pave the way for more efficient and accurate nursing documentation.

Methods:

This study was carried out at a Japanese university hospital, investigating intraoperative documentation in eight cases of PTCD and TACE. Events were recorded using both an existing voice dialogue system and traditional handwritten methods. Comparative analysis was performed to evaluate the number of events captured by each method. Surveys and structured interviews were conducted with nurses to gather qualitative insights into the system’s usability, practicality, and encountered challenges. To ensure patient safety and minimize workflow disruptions, the experimental setup was carefully designed to integrate devices seamlessly into the clinical environment while respecting existing protocols.

Results:

Speech input demonstrated effectiveness in capturing 50% to 100% of preoperative events and 40% to 100% of intraoperative events. However, its application in the postoperative phase was less effective, recording only 0% to 12% of events. Postoperative challenges were attributed to increased workload, technical issues, the need for repeated clarifications, and communication difficulties with both patients and staff while using the system. Despite these limitations, speech input showed potential for improving documentation efficiency when appropriately implemented.

Conclusions:

Key factors for successful application of speech input include ensuring patient privacy, starting with simple records, and incorporating backup documentation methods. Experiments conducted in real clinical settings, rather than simulations, emphasized the importance of designs that consider nurse-system interaction and the surrounding environment. While significant challenges remain, the findings highlight the potential of speech input technology to enhance nursing documentation efficiency and accuracy through optimized system designs and task selection. With continued refinement, this technology could reduce nurse workload and improve care quality.