Scholarlink

Purpose <p>The in-hospital incidence and disease burden of acute kidney injury (AKI) are increasing, especially among elderly patients in intensive care units (ICUs). We aim to leverage machine learning (ML) to construct and evaluate early prediction models for AKI in elderly ICU patients. AKI predictive models facilitate timely clinical decisions and ultimately improve patient outcomes.</p> Methods <p>To obtain datasets, we extracted data from the Medical Information Mart for Intensive Care IV (MIMIC-IV) from elderly patients who developed AKI after ICU admission. Three machine learning algorithms, including the neural network (NN), decision tree (DT), and extreme gradient boosting (XGBoost) algorithms, were constructed. Additionally, model performance was evaluated using the area under the receiver operating characteristic curve (AUC-ROC), and the optimal model was selected.</p> Results <p>A total of 15,316 elderly patients admitted to the ICU were included in this study, of whom 5544 (36.19%) were AKI patients. The machine learning XGBoost model outperformed the other models and achieved a superior AU-ROC (0.898). Urine volume (UV), serum creatinine (Scr) level, white blood cell count (Wbc), lactate (Lac) level, and blood urea nitrogen (Bun) level were critical predictors for the XGBoost model.</p> Conclusions <p>In this work, we developed a prediction model based on supervised machine learning to identify elderly patients with AKI progression in the ICU. Compared with other machine learning algorithms, the XGBoost model demonstrates superior predictive performance. The potential application of machine learning in daily clinical practice is demonstrated in our study.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Prediction of elderly acute kidney injury (AKI) in intensive care units (ICU) based on machine learning model

  • 期刊论文

  • 发表日期：

    2025年11月3日

查看全文
  • Yanhong Du,
  • Pingping Sun,
  • Jingyi Sheng,
  • Hongri Wang,
  • Juan Lei,
  • Wen Zhang,
  • Jiayu Song

摘要

Purpose

The in-hospital incidence and disease burden of acute kidney injury (AKI) are increasing, especially among elderly patients in intensive care units (ICUs). We aim to leverage machine learning (ML) to construct and evaluate early prediction models for AKI in elderly ICU patients. AKI predictive models facilitate timely clinical decisions and ultimately improve patient outcomes.

Methods

To obtain datasets, we extracted data from the Medical Information Mart for Intensive Care IV (MIMIC-IV) from elderly patients who developed AKI after ICU admission. Three machine learning algorithms, including the neural network (NN), decision tree (DT), and extreme gradient boosting (XGBoost) algorithms, were constructed. Additionally, model performance was evaluated using the area under the receiver operating characteristic curve (AUC-ROC), and the optimal model was selected.

Results

A total of 15,316 elderly patients admitted to the ICU were included in this study, of whom 5544 (36.19%) were AKI patients. The machine learning XGBoost model outperformed the other models and achieved a superior AU-ROC (0.898). Urine volume (UV), serum creatinine (Scr) level, white blood cell count (Wbc), lactate (Lac) level, and blood urea nitrogen (Bun) level were critical predictors for the XGBoost model.

Conclusions

In this work, we developed a prediction model based on supervised machine learning to identify elderly patients with AKI progression in the ICU. Compared with other machine learning algorithms, the XGBoost model demonstrates superior predictive performance. The potential application of machine learning in daily clinical practice is demonstrated in our study.