Background Frailty significantly complicates clinical outcomes in older adults with acute myocardial infarction, yet its progression is dynamic and heterogeneous. This study aimed to identify distinct frailty trajectory patterns and their predictors using a machine learning approach, to support evidence-based nursing interventions. Methods A prospective cohort study was conducted, enrolling 583 older adults with acute myocardial infarction hospitalized between March 2023 and March 2024. We collected multidimensional clinical, physiological, psychological, and functional data at six time points over a one-year follow-up period. A patient similarity network was constructed from these longitudinal data, and the Structural Entropy Clustering algorithm was employed to identify frailty trajectory groups. Group differences were analyzed using ANOVA and Tukey’s post hoc tests, while multinomial logistic regression was used to determine key predictors of trajectory membership. Results Four distinct frailty trajectories were identified: “Rapidly Worsening Frailty” (<InlineEquation ID="IEq1"> <EquationSource Format="TEX">\(n=78\)</EquationSource> </InlineEquation>, 13.4%), “Stable Non-Frail” (<InlineEquation ID="IEq2"> <EquationSource Format="TEX">\(n=261\)</EquationSource> </InlineEquation>, 44.7%), “Slowly Progressive Frailty” (<InlineEquation ID="IEq3"> <EquationSource Format="TEX">\(n=218\)</EquationSource> </InlineEquation>, 37.4%), and “Improving Frailty” (<InlineEquation ID="IEq4"> <EquationSource Format="TEX">\(n=26\)</EquationSource> </InlineEquation>, 4.5%). Significant differences were observed among the groups in functional status, psychological scores, nutritional status, left ventricular ejection fraction, and Charlson Comorbidity Index (<InlineEquation ID="IEq5"> <EquationSource Format="TEX">\(p<0.05\)</EquationSource> </InlineEquation>). Multivariate analysis revealed that lower functional status (Modified Barthel Index per 10-point decrease: <InlineEquation ID="IEq6"> <EquationSource Format="TEX">\(OR=9.34\)</EquationSource> </InlineEquation>, 95% CI: 7.37–11.82, <InlineEquation ID="IEq7"> <EquationSource Format="TEX">\(p<0.05\)</EquationSource> </InlineEquation>) and advanced age (<InlineEquation ID="IEq8"> <EquationSource Format="TEX">\(OR=1.07\)</EquationSource> </InlineEquation>, <InlineEquation ID="IEq9"> <EquationSource Format="TEX">\(p<0.05\)</EquationSource> </InlineEquation>) were strong predictors for the “Rapidly Worsening Frailty” trajectory, while psychological factors including anxiety (<InlineEquation ID="IEq10"> <EquationSource Format="TEX">\(OR=2.33\)</EquationSource> </InlineEquation>, <InlineEquation ID="IEq11"> <EquationSource Format="TEX">\(p<0.05\)</EquationSource> </InlineEquation>) and depression (<InlineEquation ID="IEq12"> <EquationSource Format="TEX">\(OR=2.50\)</EquationSource> </InlineEquation>, <InlineEquation ID="IEq13"> <EquationSource Format="TEX">\(p<0.05\)</EquationSource> </InlineEquation>) were significant predictors for the “Slowly Progressive Frailty” trajectory. Conclusions Frailty progression following acute myocardial infarction is heterogeneous, and distinct trajectory patterns can be identified using structural entropy clustering. These findings may support the development of differentiated nursing strategies for early identification of high-risk individuals, pending validation in multicenter settings.

Identifying frailty trajectories in older patients with acute myocardial infarction using structural entropy clustering: a prospective cohort study

期刊论文
出版模式：

开放获取
发表日期：

2026年5月14日

查看全文

Tongtong Zhang,
Haoran Yang,
Xiaoping Lou,
Chao Lan,
Naifu Tang,
Bo Li

摘要

Background

Frailty significantly complicates clinical outcomes in older adults with acute myocardial infarction, yet its progression is dynamic and heterogeneous. This study aimed to identify distinct frailty trajectory patterns and their predictors using a machine learning approach, to support evidence-based nursing interventions.

Methods

A prospective cohort study was conducted, enrolling 583 older adults with acute myocardial infarction hospitalized between March 2023 and March 2024. We collected multidimensional clinical, physiological, psychological, and functional data at six time points over a one-year follow-up period. A patient similarity network was constructed from these longitudinal data, and the Structural Entropy Clustering algorithm was employed to identify frailty trajectory groups. Group differences were analyzed using ANOVA and Tukey’s post hoc tests, while multinomial logistic regression was used to determine key predictors of trajectory membership.

Results

Four distinct frailty trajectories were identified: “Rapidly Worsening Frailty” ( \(n=78\) , 13.4%), “Stable Non-Frail” ( \(n=261\) , 44.7%), “Slowly Progressive Frailty” ( \(n=218\) , 37.4%), and “Improving Frailty” ( \(n=26\) , 4.5%). Significant differences were observed among the groups in functional status, psychological scores, nutritional status, left ventricular ejection fraction, and Charlson Comorbidity Index ( \(p<0.05\) ). Multivariate analysis revealed that lower functional status (Modified Barthel Index per 10-point decrease: \(OR=9.34\) , 95% CI: 7.37–11.82, \(p<0.05\) ) and advanced age ( \(OR=1.07\) , \(p<0.05\) ) were strong predictors for the “Rapidly Worsening Frailty” trajectory, while psychological factors including anxiety ( \(OR=2.33\) , \(p<0.05\) ) and depression ( \(OR=2.50\) , \(p<0.05\) ) were significant predictors for the “Slowly Progressive Frailty” trajectory.

Conclusions

Frailty progression following acute myocardial infarction is heterogeneous, and distinct trajectory patterns can be identified using structural entropy clustering. These findings may support the development of differentiated nursing strategies for early identification of high-risk individuals, pending validation in multicenter settings.