Integration of machine learning and large language models for screening and identifying key risk factors of acute kidney injury after cardiac surgery

Aug 5, 2026

Study aims to identify critical risk factors for acute kidney injury (AKI) after cardiac surgery by combining MIMIC-IV patient data with machine learning and large language models (LLMs). ICU data were used to select predictive features through Lasso regression and random forest, with performance assessed by 10-fold cross-validation using accuracy, sensitivity, specificity, and AUC. LLMs simulated cardiology and nephrology expert judgment to enhance clinical relevance and were validated through clinical expert discussions, yielding 18 key postoperative AKI predictors.

LanguageEnglish

CategoryHealthcare

FormatPDF

Pages13

Reading Time33 min

AccessFree to download

Objectives

Methods

Data source and study population

Feature selection and evaluation

LLM-simulated expert validation

Results

Conclusion

Keywords

TYPE Original Research PUBLISHED 06 November 2025 DOI 10.3389/fmed.2025.1618222 OPEN ACCESS EDITED BY Rajendra Bhimma, University of KwaZulu-Natal, South Africa REVIEWED BY Vikram Sabapathy, University of Virginia, United States Masao Iwagami, University of Tsukuba, Japan *CORRESPONDENCE Tao Liu [liutaocreate@gmail.com](liutaocreate@gmail.com) RECEIVED 25 April 2025 ACCEPTED 23 September 2025 PUBLISHED 06 November 2025 CITATION Li Z, Wang L, Zhang X, Wu A and Liu T (2025) Integration of machine learning and large language models for screening and identifying key risk factors of acute kidney injury after cardiac surgery. Front. Med. 12:1618222 . doi: 10.3389/fmed.2025.1618222 COPYRIGHT © 2025 Li, Wang, Zhang, Wu and Liu. This isan open-access article distributed under the terms of the Creative Commons Attribution License (CC BY) . The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms. Integration of machine learning and large language models for screening and identifying key risk factors of acute kidney injury after cardiac surgery Zishan Li1 , Lei Wang2 , Xunying Zhang1,3 , Aiping Wu4 and Tao Liu1* 1 College of Science, North China University of Science and Technology, Tangshan, China, 2 Department of Urology, North China University of Science and Technology Afﬁliated Hospital, Tangshan, China, 3 College of Automotive Engineering, Hebei College of Science and Technology, Tangshan, China, 4 School of Basic Medical Sciences, North China University of Science and Technology, Tangshan, China Objectives: This study aimed to identify critical risk factors for acute kidney injury (AKI) following cardiac surgery. By integrating patient data from the MIMIC-IV database with large language models (LLMs) and machine learning algorithms, we ensured the clinical relevance of the selected risk factors, providing robust insights for the early identiﬁcation and intervention of postoperative AKI. Methods: Intensive care unit (ICU) data of patients from the MIMIC-IV database undergoing cardiac surgery were analyzed. Lasso regression and random forest algorithms were used to select signiﬁcant predictive features from high-dimensional data. Model evaluation involved 10-fold cross-validation and metrics including accuracy, sensitivity, speciﬁcity, and the area under the curve. To enhance clinical relevance, LLMs-simulated expert judgment in cardiology and nephrology, which was further validated through discussions with clinical experts. Results: In the cohort consisting of 4,565 patients, a total of 113 important and shared risk factors for AKI were identiﬁed, including variables such as anion gap, arterial partial pressure of oxygen (PaO 2), and fraction of inspired oxygen (FiO 2) . Among these, 18 key variables were identiﬁed as postoperative AKI predictors via machine learning and LLMs-simulated expert validation. These included anchor age, Creatinine (serum), BUN (Blood Urea Nitrogen), Potassium (serum), Sodium (serum), Lactic Acid, Troponin-T, Furosemide (Lasix), Vancomycin (Random), Gentamicin (Trough), Albumin 5%, ART BP Mean, Cardiac Output (thermodilution), Brain Natriuretic Peptide (BNP), Absolute Count - Lymphs, Absolute Count - Monos, and Absolute Count - Neuts. The integration of LLMs with machine learning algorithms proved effective in accurately identifying clinically relevant risk factors. Conclusion: The proposed risk prediction approach for postoperative AKI following cardiac surgery, based on the collaborative analysis of machine learning and large language models (LLMs), effectively identiﬁed and validated key clinical risk factors. By simulating expert clinical reasoning, the LLMs signiﬁcantly e

FAQs

What is the primary objective of this study?

To identify critical risk factors for acute kidney injury following cardiac surgery using patient data combined with machine learning and large language models.

How were predictive features selected and evaluated?

Lasso regression and random forest selected significant features from high-dimensional ICU data, and model performance was measured with 10-fold cross-validation using accuracy, sensitivity, specificity, and AUC.

How did large language models improve clinical relevance in the workflow?

LLMs simulated expert judgment in cardiology and nephrology to support feature selection, and the resulting key predictors were further validated through discussions with clinical experts.

Aug 5, 2026

Upload by Genevieve