Scientists say AI like ChatGPT and Gemini can make personalized medical nutrition more accessible

A study indicates that large language models (LLMs) can provide personalized dietary advice for breast cancer patients. Such tailored diets can increase health equity for patients with variable food access. 

Researchers from Thomas Jefferson University, US, and dietitians evaluated the performances of ChatGPT and Gemini, which generated grocery lists and meal plans based on location, culture, and budget. However, the AI did not factor in patients’ age, disease stage, comorbidities, or dietary guidelines. 

“Gemini provided more comprehensive responses, including visuals and specific prices. While the dietitian-generated diets offered more adherent total daily calorie contents to the US Department of Agriculture’s estimated calorie needs, ChatGPT and Gemini offered more adherent macronutrient ratios to the Acceptable Macronutrient Distribution Range,” reads the paper.

“Overall, the meal plans were not significantly different between the LLMs and dietitians.”

Comparing LLM and dietitian plans

Researchers created 31 prompt templates to evaluate the dietary recommendations generated by ChatGPT and Gemini. Variables include cancer stage, comorbidity, location, culture, age, dietary guidelines, budget, and store.

Four board-certified oncology dietitians were asked to respond to seven of the prompts. 

The researchers evaluated responses based on nutritional content and qualitative observations, such as comparing calories and macronutrients of the LLMs and dietitian-generated meal plans.

Weight management help is required

The researchers explain that cancer, especially hormonal cancers like prostate and breast cancer, makes it challenging to maintain a healthy weight after diagnosis. 

AI tools like ChatGPT and Gemini produced culturally and budget-sensitive meal plans that were nearly on par with those of oncology dietitians.“The weight gain is multifactorial and, in part, related to the stress of treatment, the use of steroids to minimize the toxicity of chemotherapy, and other therapies that are known to slow patients’ metabolism, such as hormone therapy. Adhering to a proper nutritional plan can improve treatment response while minimizing toxicity, reducing recurrence risk, and improving survival,” reads the paper.

“However, despite the growing body of evidence supporting weight management, many patients lack access to adequate nutritional counseling and support. Integrating dietary interventions into oncology care remains inconsistent and underutilized, partly due to systemic barriers such as insurance limitations, resource availability, and healthcare infrastructure constraints.”

Improving accessibility 

The paper, published in Nutrients, notes that physicians lack the time to provide dietary counseling during clinical visits, and medical training does not adequately equip doctors with comprehensive nutrition education. “Dietary counseling is often considered an ‘ancillary service’ rather than essential, despite its benefits for patient outcomes.”

Additionally, insurance does not cover nutritional counseling, leading to unequal access to personalized dietary guidance.

“Even in institutions with dedicated oncology dietitians, patients may receive only a single consultation or generalized dietary recommendations, not the comprehensive, personalized support necessary for long-term adherence and meaningful dietary change,” the authors critique.

“Even where dietary guidance is provided, many patients struggle to implement recommendations due to financial, geographic, and cultural barriers. The cost of specialized dietary interventions, such as high-protein supplements or organic foods, can be prohibitive, particularly for anyone facing financial hardship.”

Despite visual aids and pricing from Gemini, neither LLM accounted for clinical factors like cancer stage or comorbidities.Therefore, the researchers posit LLMs as a promising tool for delivering tailored dietary recommendations to cancer patients.

Medical diets

In other news, US Food is Medicine Institute researchers predicted that nationwide implementation of medically tailored meals could save around US$32.1 billion in healthcare costs across 49 states in the first year alone. 

Recent studies from Germany and Italy have revealed significant gaps in patient nutritional care, which may impact their ability to fight disease.

A Stanford Medicine study suggests that AI-powered IV nutrition recommendations may improve care for premature infants, save time, and reduce errors. The research suggests that algorithms can read data from preemies’ electronic medical records to predict what and how many nutrients they would need.

Nutrition Insight discussed the potential of AI in nutrition with AIFS director Ilias Tagkopoulos, Ph.D., professor of computer science at the UC Davis Genome Center.