Evaluating the Effectiveness of AI-assisted Food Recommendation System on Personalized Nutrition

Last registered on December 15, 2025

Pre-Trial

Trial Information

General Information

Title
Evaluating the Effectiveness of AI-assisted Food Recommendation System on Personalized Nutrition
RCT ID
AEARCTR-0015970
Initial registration date
September 26, 2025

Initial registration date is when the trial was registered.

It corresponds to when the registration was submitted to the Registry to be reviewed for publication.

First published
October 01, 2025, 7:39 AM EDT

First published corresponds to when the trial was first made public on the Registry after being reviewed.

Last updated
December 15, 2025, 2:50 PM EST

Last updated is the most recent time when changes to the trial's registration were published.

Locations

Region

Primary Investigator

Affiliation
TAMU

Other Primary Investigator(s)

PI Affiliation
Texas A&M University
PI Affiliation
Texas A&M University
PI Affiliation
Texas A&M University
PI Affiliation
Texas A&M University

Additional Trial Information

Status
In development
Start date
2025-10-06
End date
2026-01-12
Secondary IDs
Prior work
This trial does not extend or rely on any prior RCTs.
Abstract
Nutritional needs vary for each person, shaped by their genetic profile, metabolic processes, and microbiome composition. Recent technological advances have stimulated innovative solutions for personalized dietary guidance, with Food Recommendation Systems (FRS) leveraging AI to provide tailored recommendations based on users' preferences and health goals. Yet, the current FRS is limited in integrating health profiles, preferences, and nutrition needs. To address this, the project aims to develop and evaluate the impact of applying an AI and natural language processing (NLP)-driven FRS that personalizes dietary suggestions based on individual health parameters, nutritional goals, and preferences. The impact of this project relies on improving dietary adherence, enhancing user satisfaction, and addressing specific health conditions. Our proposed AI-enabled system will assist users in recommending nutritionally customized foods while also considering suitable options for users with allergies and health concerns. The proposed method utilizes a novel deep generative network architecture to provide personalized recommendations to users based on their health profiles and preferences (precision nutrition). The system will promote healthier eating habits and address diet-related health issues through evidence-based personalized food prescriptions. Our approach is summarized as follows: First, we will utilize two publicly accessible datasets: the nutrition-based food dataset and the person's health profile and preference dataset. The first contains nutritional values (e.g., protein, carbohydrates, fat, fiber, and energy). The second contains information on age, body mass index, basal metabolic rate, targeted energy intake, physical activity level (PAL), and the existence of disease/deficiency (cardiovascular, diabetes, and iron) and food/taste preferences. Second, data will be normalized and fed to the variational autoencoder network, capturing informative features about the users' dietary needs. An NLP algorithm will be trained to adapt to user preferences. A recurrent neural network will generate recommendations to ensure the energy and nutrients align with the user's needs. Loss functions will be deployed to align our system with trustworthy AI principles by ensuring personalized food suggestions closely match expert nutritional guidelines for each user. A user interface (web or app) integrating the recommendation engine will be developed, making it a usable system with end-to-end functionality. Finally, we will implement and evaluate the impact of the developed algorithm on the dietary quality of a representative sample of the US population, including vulnerable groups (healthy living). This step consists of using a randomized controlled trial (RCT) design with a treated group (i.e., those who receive personalized dietary recommendations prescribed by the developed AI system) and a control group (those who receive standard dietary recommendations). Relevance to the IHA Mission: This project aligns with the IHA mission by contributing to key areas such as "Reducing diet-related chronic diseases," "Lowering health care costs," and "Eliminating diet-related health inequities." Through the development and evaluation of personalized dietary recommendations generated by AI, this project directly supports the IHA' s goal of integrating innovative technologies into health-promoting solutions. Further, the project has the potential to bridge the gap between responsible agriculture, healthy living, and public health outcomes, fostering collaborations across disciplines to enhance human health. In the future, we will aim to make prescriptions cost-efficient and, therefore, affordable (responsive agriculture).
External Link(s)

Registration Citation

Citation
Melo, Grace et al. 2025. "Evaluating the Effectiveness of AI-assisted Food Recommendation System on Personalized Nutrition." AEA RCT Registry. December 15. https://doi.org/10.1257/rct.15970-2.0
Experimental Details

Interventions

Intervention(s)
Intervention (Hidden)
We will recruit 1,400 U.S. adults through the Forthright platform. After an eligibility screener (≥ 18 y, English-speaking, owns a smartphone, no diet restrictions), Forthright’s server-side block randomizer will assign participants 1: 1 to two parallel arms:
• Treatment arm – full AI-Nutrition app with real-time, personalized meal suggestions generated by AI and natural-language-processing algorithms.
• Control arm – only receives general dietary suggestions (e.g., USDA MyPlate tip sheets).
The study design is single blind from the participants’ side. Participants will be blind to allocation; they are told only that they will receive “nutrition guidance”. The control arm is not informed that another group uses an app. Forthright and our technology team will see and handle the randomization and allocations, while the data analysis team will be masked to the arm codes.
We will collect dietary intake along with several behavioral questions at the baseline (Week 0, before app access) and the post-intervention point (Week 12). The dietary intake will be collected with the ASA24 24-hour recall, and the behavioral questions will be given to study participants to assess their eating habits. The treatment arm participants will be instructed to use the app at least once per week and indicate whether they followed each personalized recommendation. App analytics (log-ins, screen views, session length) will be logged automatically for subsequent engagement analyses.
Intervention Start Date
2025-10-13
Intervention End Date
2026-01-05

Primary Outcomes

Primary Outcomes (end points)
The primary outcome we are going to monitor is the difference between the HEI-2020 score at baseline (week 0) and week 12. HEI- 2020 measures the dietary quality compared with the key dietary recommendations and dietary patterns by the Dietary Guidelines for Americans. The scores range from 0 to 100, with higher values indicating closer adherence to the guideline. We will also monitor the changes in different categories, such as changes in the fruit component and the vegetable component via the HEI-2020 Fruit sub-score and the HEI-2020 Vegetable sub-score. App engagement data and behavioral questions will serve as the explanatory variables in the study.
Primary Outcomes (explanation)
Dietary Quality: Change in Healthy Eating Index-2020 (HEI-2020) score from baseline (Week 0) to end of intervention (Week 12), computed from two 24-hour dietary recalls delivered in Forthright using the USDA Automated Multiple-Pass Method (ASA24) .
Adherence: Proportion of AI-recommended meals actually consumed over the 12-week period, as recorded in the FRS app’s daily meal-tracking logs.

Secondary Outcomes

Secondary Outcomes (end points)
The secondary outcomes (HEI subscores) will also be analyzed using identical models. The behavioral variables will be employed as the explanatory variables in the models. We will also conduct heterogeneous effects based on specific groups such as age, gender, and BMI categories, and nutrition habits based on our survey questions. The diff-in-diff model will also be conducted within these groups. The results will be interpreted as exploratory.
Secondary Outcomes (explanation)

Experimental Design

Experimental Design
We will recruit 1,400 U.S. adults through the Forthright platform. After an eligibility screener (≥ 18 y, English-speaking, owns a smartphone, no diet restrictions), Forthright’s server-side block randomizer will assign participants 1: 1 to two parallel arms:
• Treatment arm – full AI-Nutrition app with real-time, personalized meal suggestions generated by AI and natural-language-processing algorithms.
• Control arm – only receives general dietary suggestions (e.g., USDA MyPlate tip sheets).
The study design is single blind from the participants’ side. Participants will be blind to allocation; they are told only that they will receive “nutrition guidance”. The control arm is not informed that another group uses an app. Forthright and our technology team will see and handle the randomization and allocations, while the data analysis team will be masked to the arm codes.
We will collect dietary intake along with several behavioral questions at the baseline (Week 0, before app access) and the post-intervention point (Week 12). The dietary intake will be collected with the ASA24 24-hour recall, and the behavioral questions will be given to study participants to assess their eating habits. The treatment arm participants will be instructed to use the app at least once per week and indicate whether they followed each personalized recommendation. App analytics (log-ins, screen views, session length) will be logged automatically for subsequent engagement analyses.
Data-collection schedule
Study week Assessment Mode Approx. time
0 (baseline) ASA24® 24-h recall; demographics; eating-behavior questionnaire Forthright survey link 20 min
1 – 12 App use (Treatment arm) + engagement logging In-app 5 min per use
12 (post) ASA24® 24-h recall; user-experience items Forthright survey link 20 min
Experimental Design Details
Randomization Method
Forthright algorithm
Randomization Unit
Two arms are included: (1) Control – intake tracking app and general advice, and (2) Treatment – intake tracking app plus AI-generated dietary advice.
Was the treatment clustered?
No

Experiment Characteristics

Sample size: planned number of clusters
Primary design is individual-level randomization via Forthright. If operational constraints require batching or geographic assignment that induces clustering, we will amend the registry to specify the cluster unit and count and will use cluster-robust SEs accordingly.
Sample size: planned number of observations
700 U.S. adults per arm (1,400 U.S. adults in total ) through the Forthright platform.
Sample size (or number of clusters) by treatment arms
Sample size: N≈700 for treatment group
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
0.2
IRB

Institutional Review Boards (IRBs)

IRB Name
Texas A&M University HUMAN RESEARCH PROTECTION PROGRAM
IRB Approval Date
2025-12-11
IRB Approval Number
STUDY2025-1209
Analysis Plan

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information

Post-Trial

Post Trial Information

Study Withdrawal

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information

Intervention

Is the intervention completed?
No
Data Collection Complete
Data Publication

Data Publication

Is public data available?
No

Program Files

Program Files
Reports, Papers & Other Materials

Relevant Paper(s)

Reports & Other Materials