JMIR Publications

ABSTRACT

Background:

Long-term care (LTR) faces critical challenges driven by workforce shortages, an aging population, and a growing population of people living with dementia. Administrative burdens add to this pressure, as healthcare professionals spend up to 40% of their working time on administration and documentation. Speech-driven AI reporting (SDR) may offer a technological solution to alleviate administrative reporting workload and enhance the workflow efficiency of care workers.

Objective:

This study aimed to empirically study the effects of SDR on documentation time, transcription accuracy measured by Word Error Rate, user experiences, and the client-caregiver interaction within nursing homes and home care settings.

Methods:

A mixed-methods study, involving 21 healthcare organizations, was conducted in the Netherlands between January and September of 2025. An experimental evaluation study comparing speech-driven and typed reporting under controlled conditions (n=35), complemented by a cross-sectional questionnaire study among care professionals from 14 elderly care organizations (n=293). Documentation time and Word Error Rate were analyzed using linear mixed models. Associations between system use duration and user experience were examined using correlation analyses.

Results:

The controlled evaluation study demonstrated a significant reduction in reporting time. SDR was found to be significantly faster than typing (p < 0.01), with a significant interaction between reporting device and method (p = 0.01), being 3.5 times faster on smartphones (34 seconds vs. 122 s) and 2.3 times faster on laptops (43 vs. 102 seconds). The SDR AI software demonstrated high transcription accuracy (Word Error Rate <0.05). SDR did change the reporting process: healthcare workers reported more directly after they provided care for their clients (19.0% vs 42.1%; p<0.001) and fewer reports were made after the end of their shift. Also, no correlations between SDR use and technology acceptance aspects, or perceived work pressure were determined.

Conclusions:

The current SDR technology offers time savings and high accuracy regardless of the device used (smartphone or laptop). However, the technological capability alone does not automatically translate to reduced perceived work pressure by care workers. The findings suggest that the challenge has shifted from technical feasibility to implementation strategy and behavioral change.