Virginia Via Research Day Book 2026

Medical Student Research Education and Simulation

09 EVALUATING THE ACCURACY OF ARTIFICAL INTELLGIENCE GENERATED NUTRITION PLANS COMPARED TO EVIDENCE BASED MEDICAL NUTRITION THERAPY GUIDELINES

Meghan Wilson, PhD; Manisha Seelam, OMS III; Lauren Provinsal, OMS III; Michella Donfack, OMS III; Mehar Nasir, OMS III Corresponding author: mseelam@vt.vcom.edu

VCOM-Virginia, Blacksburg, Virginia

Traditional medical nutrition therapy (MNT) is a key component of chronic disease management and has been shown to improve clinical outcomes across a variety of conditions. Traditional MNTs are grounded in scientific evidence, including the ketogenic diet for epilepsy, the Mediterranean-DASH intervention Neurodegenerative Delay (MIND) diet for Alzheimer’s disease, anti-inflammatory diets for multiple sclerosis, and healthy maternal dietary patterns for pre-eclampsia.

ChatGPT, a widely used artificial intelligence (AI) platform, is increasingly being used in healthcare because its easy accessibility allows both the public and healthcare professionals to rapidly obtain and reference information. However, it is still unclear whether AI-generated nutritional recommendations align with validated clinical guidelines or MNT. This study will use ChatGPT to generate personalized

nutrition plans and compare them with traditional, evidence-based MNT for epilepsy, multiple sclerosis, Alzheimer’s disease, and pre-eclampsia. Each plan will be evaluated for nutrient composition, caloric distribution, and adherence to disease-specific recommendations. We predict that AI-generated MNT will be broadly comparable to guideline-based plans in overall structure while potentially differing in disease

Table of Contents

165 2026 Research Recognition Day