Mortality risk assessment and interpretability analysis of preterm infants in the NICU using machine learning models

Corresponding author: WU Xiayang, 56425477@qq.com

DOI: 10.12201/bmr.202503.00066

•  

  Abstract: Objective: Aimed at using machine learning algorithms to predict the risk of neonatal ICU mortality, providing clinicians with an early diagnosis and risk assessment tool to assist in decision-making. Methods: Retrospectively collecting clinical data of preterm infants from the PIC database. Cases were divided into mortality and survival groups based on ICU outcomes. Key clinical characteristics potentially affecting preterm infant outcomes were screened using LASSO regression analysis and multivariate logistic regression analysis. The study balanced the data using the SMOTE algorithm and constructed predictive models using seven machine learning models (e.g., LightGBM, random forest), evaluating their performance. Model interpretation was performed using the Shapley Additive Explanations (SHAP) algorithm. Results: A total of 923 infants were included in the final analysis. The survival group comprised 886 infants, and the death group comprised 37 infants. A total of 38 clinical characteristics were collected. LASSO screening identified 8 variables significantly associated with neonatal ICU mortality, including lactate, chloride concentration, neutrophils, and red blood cell distribution width. Multivariate logistic regression analysis revealed that lactate and respiratory rate were independent predictors of neonatal ICU outcomes. The LightGBM model achieved an AUC of 0.972 and outperformed other models in terms of accuracy and precision. Furthermore, SHAP analysis enhanced model interpretability. The results indicated that respiratory rate and lactate contributed most significantly to the prediction of infant mortality risk. Conclusion: This study provides reliable tools for early identification and intervention of preterm infant outcomes, emphasizing the importance of key physiological indicators. Future multi-center data validation is needed to enhance the models generalizability and further optimize algorithm performance.

  Key words: Premature infants, ICU mortality risk, machine learning, LightGBM model, risk prediction

  Submit time: 21 March 2025

  Copyright: The copyright holder for this preprint is the author/funder, who has granted biomedRxiv a license to display the preprint in perpetuity.
• html

• 图表

• zhangwei, qindawei, tianwenjun, chenling, yanmi, guxiu, fehejian, tianchangjun. Risk factors and clinical outcome of meconium-stained amniotic fluid in preterm infants. 2025. doi: 10.12201/bmr.202502.00007

  Chen Lili. Study on the immune regulatory mechanism of ω－3PUFA in premature infants with brain injury. 2024. doi: 10.12201/bmr.202408.00011

  Construction and application evaluation of risk prediction model and nomogram for shivering during cesarean sectio. 2025. doi: 10.12201/bmr.202501.00053

  wangyifan, shichaojun, maanning. Comparison of risk prediction models for atherosclerosis in type 2 diabetes mellitus. 2024. doi: 10.12201/bmr.202404.00007

  Yang ShuJie, Zeng Ying, Yi HaiYing, Huang Jie, Gan Bin, Wei Xian, Tang Yanlin. Observation of the therapeutic effect of less invasive surfactant administration on respiratory distress syndrome in extremely premature infants. 2024. doi: 10.12201/bmr.202410.00018

  周妮, GU Qiang. Risk factors and risk prediction for the course of lobar pneumonia in children. 2025. doi: 10.12201/bmr.202504.00059

  SHU Wenxiu, LUO Liufei, Tong Jiaqi, Le Jing. Construction and effect evaluation of risk prediction model for grades 3-4 myeloma bone disease in newly diagnosed multiple myeloma patients. 2024. doi: 10.12201/bmr.202410.00057

  Mo Wei, Xiang Ya, Liao Qiujiao, He Liu, Ling Chaoling, Lu Qixiang, Liu Fangyin. Research progress on risk prediction model of postoperative delirium in elderly patients with hip fractureWEI Yunshi1? MO Wei1? XIANG YA1? LIAO Qiujiao1? HE Liu2? LING Chaoling2? LU Qixiang2? LIU Fangyin3▲. 2024. doi: 10.12201/bmr.202409.00029

  lanyushan, lijiao. Machine Learning Methods for Confounding Control in Causal Inference. 2022. doi: 10.12201/bmr.202203.00015

  fengli. A Comparative Study on the Accuracy of Nine Combined Machine Learning Algorithms in Early Diagnosis of Tumors Based on High-dimensional dataFeng Li 1,*, Yue Xiaofei 2. 2021. doi: 10.12201/bmr.202108.00016

• ID	Submit time	Number	Download
  1	2025-02-19	bmr.202503.00066V1	Download

• Public  Anonymous  To author only Submit Comments