Automated Diet Capture Using Voice Alerts and Speech Recognition on Smartphones: Pilot Usability and Acceptability...

Area of Science:

• Health Informatics
• Human-Computer Interaction
• Nutrition Science

Background:

• Dietary habit monitoring is crucial for managing chronic diseases like type 2 diabetes.
• Speech recognition and natural language processing offer potential for automated dietary logging.
• Usability and acceptability of these technologies require further investigation.

Purpose of the Study:

• To explore the usability and acceptability of speech recognition and natural language processing for automated diet logging.
• To compare the effectiveness of voice versus text-based diet logging methods.

Main Methods:

• Development of base2Diet, an iOS application for voice and text food logging.
• A 28-day, two-arm, two-phase pilot study with 18 participants (9 per arm: voice vs. text).
• Participants received timed meal logging reminders, with adjustable timing in phase II.

Main Results:

• Voice logging resulted in 1.7 times more distinct diet logging events per participant (P=.03).
• Active days per participant were 1.5 times higher in the voice arm (P=.04).
• The voice arm had a significantly lower attrition rate (1 dropout) compared to the text arm (5 dropouts).

Conclusions:

• Voice technologies show significant potential for automated diet capture via smartphones.
• Voice-based diet logging is more effective and better received by users than text-based methods.
• Findings support further research and development of accessible tools for dietary monitoring and health promotion.