Using artificial intelligence models to generate dietary recommendations for chronic kidney disease patients: A comparative cross-sectional study

IF 7.9 2区医学 Q1 NUTRITION & DIETETICS

Clinical nutrition Pub Date : 2025-12-01 Epub Date: 2025-10-30 DOI:10.1016/j.clnu.2025.10.014

Binyu Yang , Wei Wei , Caihong Liu , Yongxiu Huang , Jinglei Ren , Ya Yuan , Ping Fu , Yuliang Zhao

{"title":"Using artificial intelligence models to generate dietary recommendations for chronic kidney disease patients: A comparative cross-sectional study","authors":"Binyu Yang , Wei Wei , Caihong Liu , Yongxiu Huang , Jinglei Ren , Ya Yuan , Ping Fu , Yuliang Zhao","doi":"10.1016/j.clnu.2025.10.014","DOIUrl":null,"url":null,"abstract":"<div><h3>Background & aims</h3><div>Artificial intelligence (AI) models are increasingly being used to assist in chronic kidney disease (CKD) dietary guidance, but concerns about their accuracy remain. This study aimed to assess the performance of three AI models (ChatGPT, DeepSeek, Gemini) in accordance with the Kidney Disease: Improving Global Outcomes (KDIGO) 2024 Clinical Practice Guideline for the Evaluation and Management of Chronic Kidney Disease.</div></div><div><h3>Methods</h3><div>We compared ChatGPT, DeepSeek, and Gemini with the KDIGO 2024 guidelines through expert and cross-AI assessments. Outputs were scored across five domains: accuracy, clarity, completeness, practicality, and safety.</div></div><div><h3>Results</h3><div>The AI models were able to generate well-structured dietary advice for CKD patients. DeepSeek received a higher score (7.625) for practicality compared to ChatGPT (6.5) and Gemini (7.0) (p = 0.041). ChatGPT showed numerically higher accuracy scores than the other two models, although there was no statistically significant difference (ChatGPT: 7.25, DeepSeek: 6.875, Gemini: 6.875; p = 0.715). Similarly, scores for clarity (p = 0.489), completeness (p = 0.432) and safety (p = 0.413) were not significantly different. The AI models share several common limitations: they might overlook contraindications (such as recommending a low-protein diet for patients with severe hypoalbuminemia), neglect age-specific needs, and lack real-time knowledge updates.</div></div><div><h3>Conclusions</h3><div>While AI shows clinical potential, its outputs still require expert validation, real-time updates, and enhanced safety protocols to ensure reliable clinical integration.</div></div>","PeriodicalId":10517,"journal":{"name":"Clinical nutrition","volume":"55 ","pages":"Pages 76-80"},"PeriodicalIF":7.9000,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Clinical nutrition","FirstCategoryId":"3","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S0261561425002869","RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"2025/10/30 0:00:00","PubModel":"Epub","JCR":"Q1","JCRName":"NUTRITION & DIETETICS","Score":null,"Total":0}

引用次数: 0

Abstract

Background & aims

Artificial intelligence (AI) models are increasingly being used to assist in chronic kidney disease (CKD) dietary guidance, but concerns about their accuracy remain. This study aimed to assess the performance of three AI models (ChatGPT, DeepSeek, Gemini) in accordance with the Kidney Disease: Improving Global Outcomes (KDIGO) 2024 Clinical Practice Guideline for the Evaluation and Management of Chronic Kidney Disease.

Methods

We compared ChatGPT, DeepSeek, and Gemini with the KDIGO 2024 guidelines through expert and cross-AI assessments. Outputs were scored across five domains: accuracy, clarity, completeness, practicality, and safety.

Results

The AI models were able to generate well-structured dietary advice for CKD patients. DeepSeek received a higher score (7.625) for practicality compared to ChatGPT (6.5) and Gemini (7.0) (p = 0.041). ChatGPT showed numerically higher accuracy scores than the other two models, although there was no statistically significant difference (ChatGPT: 7.25, DeepSeek: 6.875, Gemini: 6.875; p = 0.715). Similarly, scores for clarity (p = 0.489), completeness (p = 0.432) and safety (p = 0.413) were not significantly different. The AI models share several common limitations: they might overlook contraindications (such as recommending a low-protein diet for patients with severe hypoalbuminemia), neglect age-specific needs, and lack real-time knowledge updates.

Conclusions

While AI shows clinical potential, its outputs still require expert validation, real-time updates, and enhanced safety protocols to ensure reliable clinical integration.

查看原文本刊更多论文

使用人工智能模型为慢性肾病患者提供饮食建议：一项比较横断面研究

背景与目的：人工智能（AI）模型越来越多地被用于辅助慢性肾脏疾病（CKD）的饮食指导，但对其准确性仍然存在担忧。本研究旨在根据肾脏疾病：改善全球结果（KDIGO） 2024慢性肾脏疾病评估和管理临床实践指南，评估三种人工智能模型（ChatGPT, DeepSeek, Gemini）的性能。方法：通过专家评估和交叉ai评估，将ChatGPT、DeepSeek和Gemini与KDIGO 2024指南进行比较。输出在五个领域进行评分：准确性、清晰度、完整性、实用性和安全性。结果：AI模型能够为CKD患者提供结构良好的饮食建议。DeepSeek在实用性方面的得分（7.625）高于ChatGPT（6.5）和Gemini (7.0) （p = 0.041）。ChatGPT在数值上的准确率得分高于其他两种模型，但差异无统计学意义（ChatGPT: 7.25, DeepSeek: 6.875, Gemini: 6.875; p = 0.715）。同样，清晰度（p = 0.489）、完整性（p = 0.432）和安全性（p = 0.413）得分也无显著差异。人工智能模型有几个共同的局限性：它们可能会忽略禁忌症（例如为严重低白蛋白血症患者推荐低蛋白饮食），忽视特定年龄的需求，并且缺乏实时的知识更新。结论：虽然人工智能显示出临床潜力，但其产出仍需要专家验证、实时更新和增强的安全协议，以确保可靠的临床整合。

本文章由计算机程序翻译，如有差异，请以英文原文为准。

求助全文

约1分钟内获得全文求助全文

来源期刊

Clinical nutrition 医学-营养学

CiteScore

14.10

自引率

6.30%

发文量

356

审稿时长

28 days

期刊介绍： Clinical Nutrition, the official journal of ESPEN, The European Society for Clinical Nutrition and Metabolism, is an international journal providing essential scientific information on nutritional and metabolic care and the relationship between nutrition and disease both in the setting of basic science and clinical practice. Published bi-monthly, each issue combines original articles and reviews providing an invaluable reference for any specialist concerned with these fields.