Interactive Conversational Agents to Improve Dietary Behaviors for Health Promotion: Mixed Systematic Review
Bibliographic record
Abstract
Background: Chronic diseases are the leading global cause of death, largely driven by Western lifestyles characterized by poor diets and physical inactivity. Digital interventions offer promising tools to support health behavior change. Interactive conversational agents (CAs) provide real-time, personalized meal planning and dietary advice. Their interactive nature and adaptability make them valuable for promoting healthy dietary behaviors in the context of diet-related chronic diseases. However, evidence of their effectiveness remains limited. Systematic evaluations of their impact, features, and user acceptability are needed to clarify their role in public health strategies for improving dietary behaviors and preventing chronic diseases. Objective: This review aimed to evaluate the effectiveness of CAs in improving dietary behaviors, to describe their features, functions, conversational capabilities, and impact on nutritional knowledge, usability, acceptability, user experience, and engagement. Methods: Five electronic databases were searched: MEDLINE, CINAHL, Embase, Web of Science, and PsycINFO. We only included sources that focused on the use of CAs to change dietary behavior. Eligible studies were published since 2013 in English, French, or Spanish. Two independent reviewers screened studies, with a third resolving disagreements. The quality of studies was appraised using the Mixed Methods Appraisal Tool (McGill University). Quantitative and qualitative findings were synthesized narratively. Results: In total, 2200 references were identified, and after screening and eligibility assessment, 11 references (10 studies with approximately 20-480 participants) were included. Among the included studies, improvements in fruit and vegetable intake were reported in 2 studies (P=.04 and P=.005). One study found significant increases in adherence to the Mediterranean diet at 6 weeks with gains maintained at 12 weeks. Two additional studies reported enhanced nutritional knowledge (eg, nutrition label use). Effects on protein, whole grains, sugar, sodium, and caffeine intake were mixed or nonsignificant. Some studies reported increased physical activity (+109.8 min/wk) and reduced alcohol use for stress management. One randomized controlled trial showed modest but significant weight loss and decrease in waist circumference (-2.1 cm, 95% CI -3.5 to 0.7; P=.003). Engagement varied between studies. Usability and user experience were generally positive; goal setting, feedback, and tailored recommendations were linked to higher satisfaction. Reported challenges included unnatural conversation style, simplistic content, and limited perceived usefulness. Conclusions: CAs show promising potential to improve dietary behaviors, with evidence of gains in fruit and vegetable intake, Mediterranean diet adherence, nutritional knowledge, physical activity, and modest weight loss. Overall usability was favorable, but variability in outcomes, high attrition, and limited impact on social support highlight areas for refinement. Future research should use larger samples, longer follow-up, standardized outcomes, and strategies to enhance sustained engagement and inclusivity. Systematic evaluations and refined designs are essential to establish the role of CAs as scalable, evidence-based tools in chronic disease prevention.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.015 | 0.060 |
| Meta-epidemiology (narrow) | 0.002 | 0.001 |
| Meta-epidemiology (broad) | 0.008 | 0.008 |
| Bibliometrics | 0.010 | 0.012 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.004 | 0.003 |
| Open science | 0.003 | 0.002 |
| Research integrity | 0.002 | 0.002 |
| Insufficient payload (model declined to judge) | 0.008 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".