Author Correction: A Machine-learning Approach to Forecast Aggravation Risk in Patients with Acute Exacerbation of Chronic Obstructive Pulmonary Disease with Clinical Indicators

Junfeng Peng, Chuan Chen, Mi Ζhou, Xiaohua Xie, Yuqi Zhou, Ching‐Hsing LuoScientific 2021 · DOI

Abstract

This article presents an author correction for the study "A Machine-learning Approach to Forecast Aggravation Risk in Patients with Acute Exacerbation of Chronic Obstructive Pulmonary Disease with Clinical Indicators," originally published in Scientific Reports. The correction addresses inaccuracies in the baseline clinical data reported for the patient cohorts, ensuring a rigorous foundation for the machine-learning models developed to predict patient outcomes.

Introduction

Patients with Acute Exacerbation of Chronic Obstructive Pulmonary Disease (AECOPD) present a significant clinical challenge due to the variability in their progression. While many stabilize with standard treatment, a subset of patients experiences "aggravation"—a rapid decline requiring intensive medical intervention.

Early identification of these high-risk individuals is critical for improving clinical outcomes. The original research utilized machine-learning (ML) algorithms to forecast this risk using standard clinical indicators, aiming to provide a tool for personalized and timely medical responses.

Methodology and Clinical Indicators

The predictive models were trained using a variety of clinical features collected at the time of admission. These indicators serve as the input features for the ML architecture and include:

  • Vital Signs: Temperature, pulse rate, respiratory rate, and blood pressure (systolic and diastolic).
  • Demographics: Age, sex, and smoking history.
  • Clinical History: Frequency of previous hospitalizations (NOH).
  • Comorbidities: Presence of secondary conditions such as hypertension (HTN), diabetes mellitus (DM), coronary heart disease (CHD), pulmonary heart disease (PHD), and others.

Patients were categorized into two primary groups for model training and validation: a Mild Group (low risk) and a Severe Group (high risk).

The Corrected Dataset

The study analyzed a total of 410 patient cases. The correction specifically updates the demographic and clinical distribution within these groups to reflect the true characteristics of the study population.

Table 1: Corrected clinical characteristics of the Mild (low risk) and Severe (high risk) patient groups.

As shown in the corrected Table 1, the cohort consists of:

  • Mild Group: 208 cases (50.7%), with 178 males (85.6%) and 163 patients with a smoking history (78.4%).
  • Severe Group: 202 cases (49.3%), with 163 males (80.7%) and 39 females (19.3%).

Key Insights from Clinical Data

The corrected data highlights several significant differences between the low-risk and high-risk groups, which likely served as strong predictive signals for the machine-learning models:

  1. Prior Hospitalization Frequency: The "Severe" group had a markedly higher average of prior hospitalizations (6.3 ± 7.3) compared to the "Mild" group (3.8 ± 2.8). This suggests that a history of frequent exacerbations is a primary indicator of future aggravation risk.
  2. Respiratory Strain: Patients in the high-risk group exhibited higher average respiratory rates (24.5 breaths/min) than those in the low-risk group (21.6 breaths/min), reflecting greater physiological stress.
  3. Comorbidity Burden: The prevalence of certain comorbidities was significantly higher in the Severe group. For instance, Pulmonary Heart Disease (PHD) was present in 32.2% of high-risk patients compared to 17.3% of low-risk patients. Similarly, Diabetes Mellitus (DM) was more than twice as prevalent in the Severe group (20.3% vs. 8.2%).

Conclusion

The corrections provided in this notice ensure the accuracy of the descriptive statistics for the patient population used in the study. By clarifying the "ground truth" characteristics of the Mild and Severe groups, this update reinforces the reliability of the machine-learning approach to forecasting aggravation risk in AECOPD patients.

Summary updated

Abstract

An amendment to this paper has been published and can be accessed via a link at the top of the paper.
BibTeX
@inproceedings{peng2021author,
  author = {Junfeng Peng and Chuan Chen and Mi Ζhou and Xiaohua Xie and Yuqi Zhou and Ching‐Hsing Luo},
  title = {{Author Correction: A Machine-learning Approach to Forecast Aggravation Risk in Patients with Acute Exacerbation of Chronic Obstructive Pulmonary Disease with Clinical Indicators}},
  booktitle = {Scientific},
  year = {2021},
  doi = {10.1038/s41598-020-76434-2},
}
Figure