Related Experiment Video
Updated: Aug 12, 2026
Identification of Disease-related Spatial Covariance Patterns using Neuroimaging Data
Published on: June 26, 2013
Development of a Bedside Decision Tree for Postexacerbation Risk Stratification: Translating Neuroimmune Biomarker
Cheng Chen1, Shanshan Liu2, Jian Dong1
1Department of Respiratory and Critical Care Medicine, Union Jiangbei Hospital, Huazhong University of Science and Technology, Wuhan Hubei, 430100, China, hust.edu.cn.
Background:
The acute exacerbation of chronic obstructive pulmonary disease (AECOPD) is characterized by systemic inflammatory response and neuroimmune dysregulation. The dynamic changes of certain key neuroimmune biomarkers during the course of AECOPD and their impact on long-term patient outcomes have not been thoroughly studied.
Objective:
This study aims to develop a bedside clinical decision tree based on neuroimmune biomarker dynamics to stratify post-AECOPD patient risk and guide personalized discharge planning.
Methods:
This prospective observational study included 273 patients hospitalized due to AECOPD. Serum levels of brain-derived neurotrophic factor (BDNF), programmed cell death protein 1 (PD-1), matrix metalloproteinase-9 (MMP-9), and inflammatory cytokines (IL-1β, IL-6, IL-10, and TNF-α) were measured within 24 h of admission (T1) and within 48 h after clinical stability at discharge (T2). Unsupervised clustering analysis was performed based on the dynamic changes in biomarkers, and multivariable logistic regression was used to assess their association with 90-day clinical outcomes (acute exacerbation, readmission). Additionally, a biomarker-based decision tree model was developed, and its performance was compared with traditional clinical assessment methods.
Results:
Three distinct biomarker response patterns were identified: "Coordinated Improvement " (60.0%), "Inflammatory Rebound" (27.8%), and "Poor Neuro-repair" (12.1%). Among them, the "Poor Neuro-repair" phenotype was the strongest independent predictor of acute exacerbation events at 90 days (adjusted OR 3.42, 95% CI 1.78-6.57, p < 0.001). The dynamic change in BDNF (ΔBDNF) showed predictive value for acute exacerbation events (AUC 0.84). The biomarker-based decision tree model classified patients into four risk levels, with significant differences in the 30-day event incidence (ranging from 3.6% to 68.4%). Its predictive accuracy (AUC 0.87) was markedly superior to that of clinical judgment (AUC 0.74) and the GOLD standard (AUC 0.68).
Conclusions:
The biomarker response patterns after acute exacerbation, particularly the "Poor Neuro-repair" phenotype and its dynamic changes in BDNF, are closely associated with short-term clinical outcomes. The decision tree model developed based on this information provides a preliminary approach for risk stratification and discharge planning in AECOPD patients, demonstrating superior efficacy compared to traditional clinical assessment methods.
