Establishment and validation of a machine learning model to stratify malnutrition risk in hospitalized older patients with chronic heart failure.

IF 1.6 4区医学 Q4 NUTRITION & DIETETICS

Asia Pacific journal of clinical nutrition Pub Date : 2026-02-01 DOI:10.6133/apjcn.202602_35(1).0014

Qiuhong Sun, Jing Che

{"title":"Establishment and validation of a machine learning model to stratify malnutrition risk in hospitalized older patients with chronic heart failure.","authors":"Qiuhong Sun, Jing Che","doi":"10.6133/apjcn.202602_35(1).0014","DOIUrl":null,"url":null,"abstract":"Background and objectives: Malnutrition among older hospitalized adults with chronic heart failure (CHF) is associated with adverse clinical outcomes, yet reliable early risk stratification tools remain lacking. This study aimed to develop and validate a machine learning (ML) model for malnutrition risk stratification in this population.Methods and study design: Malnutrition among older hospitalized adults with chronic heart failure (CHF) is associated with adverse clinical outcomes, yet reliable early risk stratification tools remain lacking. This study aimed to develop and validate a machine learning (ML) model for malnutrition risk stratification in this population.Results: Malnutrition prevalence was 44.1% (348/790). In the internal testing, CatBoost (CAT) achieved superior performance with an AUC of 0.901 (95% confidence interval [CI]: 0.858-0.943), accuracy of 0.840, recall of 0.753, and the lowest Brier score of 0.113. This model demonstrated strong calibration, clinical utility, and the highest composite score (62/64). External validation confirmed CAT's generalizability (AUC: 0.916, 95% CI: 0.887-0.945). SHAP analysis identified body mass index (BMI), calf circumference, New York Heart Association (NYHA) classification, age, and diabetes as signifi-cant contributors to malnutrition risk.Conclusions: The CAT-based model effectively stratifies malnutrition risk in older hospitalized CHF patients, offering a tool for early intervention to improve outcomes. Further multicenter prospective studies are needed to validate its real-world applicability.","PeriodicalId":8486,"journal":{"name":"Asia Pacific journal of clinical nutrition","volume":"35 1","pages":"143-157"},"PeriodicalIF":1.6000,"publicationDate":"2026-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12823253/pdf/","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Asia Pacific journal of clinical nutrition","FirstCategoryId":"3","ListUrlMain":"https://doi.org/10.6133/apjcn.202602_35(1).0014","RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q4","JCRName":"NUTRITION & DIETETICS","Score":null,"Total":0}

引用次数: 0

Abstract

Background and objectives: Malnutrition among older hospitalized adults with chronic heart failure (CHF) is associated with adverse clinical outcomes, yet reliable early risk stratification tools remain lacking. This study aimed to develop and validate a machine learning (ML) model for malnutrition risk stratification in this population.

Methods and study design: Malnutrition among older hospitalized adults with chronic heart failure (CHF) is associated with adverse clinical outcomes, yet reliable early risk stratification tools remain lacking. This study aimed to develop and validate a machine learning (ML) model for malnutrition risk stratification in this population.

Results: Malnutrition prevalence was 44.1% (348/790). In the internal testing, CatBoost (CAT) achieved superior performance with an AUC of 0.901 (95% confidence interval [CI]: 0.858-0.943), accuracy of 0.840, recall of 0.753, and the lowest Brier score of 0.113. This model demonstrated strong calibration, clinical utility, and the highest composite score (62/64). External validation confirmed CAT's generalizability (AUC: 0.916, 95% CI: 0.887-0.945). SHAP analysis identified body mass index (BMI), calf circumference, New York Heart Association (NYHA) classification, age, and diabetes as signifi-cant contributors to malnutrition risk.

Conclusions: The CAT-based model effectively stratifies malnutrition risk in older hospitalized CHF patients, offering a tool for early intervention to improve outcomes. Further multicenter prospective studies are needed to validate its real-world applicability.

查看原文本刊更多论文

老年慢性心力衰竭住院患者营养不良风险分层的机器学习模型的建立与验证

背景和目的：住院老年慢性心力衰竭（CHF）患者营养不良与不良临床结果相关，但仍缺乏可靠的早期风险分层工具。本研究旨在开发和验证该人群营养不良风险分层的机器学习（ML）模型。方法和研究设计：住院老年慢性心力衰竭（CHF）患者营养不良与不良临床结果相关，但仍缺乏可靠的早期风险分层工具。本研究旨在开发和验证该人群营养不良风险分层的机器学习（ML）模型。结果：营养不良患病率为44.1%（348/790）。在内部测试中，CatBoost （CAT）的AUC为0.901(95%可信区间[CI]: 0.858-0.943)，准确率为0.840，召回率为0.753，最低Brier评分为0.113。该模型具有很强的校准性、临床实用性和最高的综合评分（62/64）。外部验证证实了CAT的可推广性（AUC: 0.916, 95% CI: 0.887-0.945）。SHAP分析发现身体质量指数（BMI）、小腿围、纽约心脏协会（NYHA）分类、年龄和糖尿病是营养不良风险的重要因素。结论：基于cat的模型有效地对老年住院CHF患者的营养不良风险进行分层，为早期干预改善预后提供了工具。需要进一步的多中心前瞻性研究来验证其在现实世界中的适用性。

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

Asia Pacific journal of clinical nutrition 医学-营养学

CiteScore

2.50

自引率

7.70%

发文量

审稿时长

6-12 weeks

期刊介绍： The aims of the Asia Pacific Journal of Clinical Nutrition (APJCN) are to publish high quality clinical nutrition relevant research findings which can build the capacity of clinical nutritionists in the region and enhance the practice of human nutrition and related disciplines for health promotion and disease prevention. APJCN will publish original research reports, reviews, short communications and case reports. News, book reviews and other items will also be included. The acceptance criteria for all papers are the quality and originality of the research and its significance to our readership. Except where otherwise stated, manuscripts are peer-reviewed by at least two anonymous reviewers and the Editor. The Editorial Board reserves the right to refuse any material for publication and advises that authors should retain copies of submitted manuscripts and correspondence as material cannot be returned. Final acceptance or rejection rests with the Editorial Board