Evaluation of Dietary Management Using Artificial Intelligence and Human Interventions: Nonrandomized Controlled Trial
Bibliographic record
Abstract
BACKGROUND: There has been an increase in personal health records with the increased use of wearable devices and smartphone apps to improve health. Traditional health promotion programs by human professionals have limitations in terms of cost and reach. Due to labor shortages and to save costs, there has been a growing emphasis in the medical field on building health guidance systems using artificial intelligence (AI). AI will replace advanced human tasks to some extent in the future. However, it is difficult to sustain behavioral change through technology alone at present. OBJECTIVE: This study investigates whether AI alone can effectively encourage healthy behaviors or whether human interventions are needed to achieve and sustain health-related behavioral change. We examined the effectiveness of AI and human interventions to encourage dietary management behaviors. In addition, we elucidated the conditions for maximizing the effect of AI on health improvement. We hypothesized that the combination of AI and human interventions will maximize their effectiveness. METHODS: We conducted a 3-month experiment by recruiting participants who were users of a smartphone diet management app. We recruited 102 participants and divided them into 3 groups. Treatment group I received text messages using the standard features of the app (AI-based text message intervention). Treatment group II received video messages from a companion, in addition to the text messages (combined text message and human video message intervention by AI). The control group used the app to keep a dietary record, but no feedback was provided (no intervention). We examine the participants' continuity and the effects on physical indicators. RESULTS: Combined AI and video messaging (treatment group II) led to a lower dropout rate from the program compared to the control group, and the Cox proportional-hazards model estimate showed a hazard ratio (HR) of 0.078, which was statistically significant at the 5% level. Further, human intervention with AI and video messaging significantly reduced the body fat percentage (BFP) of participants after 3 months compared to the control group, and the rate of reduction was greater in the group with more individualized intervention. The AI-based text messages affected the BMI but had no significant effect on the BFP. CONCLUSIONS: This experiment shows that it is challenging to sustain participants' healthy behavior with AI intervention alone. The results also suggest that even if the health information conveyed is the same, the information conveyed by humans and AI is more effective in improving health than the information sent by AI alone. The support received from the companion in the form of video messages may have promoted voluntary health behaviors. It is noteworthy that companions were competent, even though they were nonexperts. This means that person-to-person communication is crucial for health interventions.
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
Direct model labels (unvalidated)
Per-model category and study-design labels from the labeling rounds. They are machine output, unvalidated, and the disagreement between models ships as data. No study design here is MEDLINE-validated yet.
| Model arm | Categories | Study design | Confidence |
|---|---|---|---|
| gemma | no category Domain: not available · Genre: Empirical About the Canadian research system: no · About a Canadian topic: no | Non-randomized trial | high |
| gpt | no category Domain: not available · Genre: Empirical About the Canadian research system: no · About a Canadian topic: no | Non-randomized trial | high |
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.014 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.001 | 0.000 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.002 | 0.000 |
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, unvalidatedLabeled directly by 2 models reading the full record.
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".