Machine Learning-Based Mortality Prediction Model for Critically Ill Cancer Patients Admitted to the Intensive Care Unit (CanICU) - 28-day Mortality Prediction

Aug 5, 2026

Upload by Patrick

Background: As cancer patients increasingly require ICU care for cancer- or treatment-related complications, accurate mortality prediction remains challenging. This study presents CanICU, a machine-learning mortality prediction model for adult critically ill cancer patients admitted to the ICU. Methods: CanICU was developed and validated using 6900 patients across three cohorts from MIMIC, Yonsei Cancer Center, and Samsung Medical Center, using nine practical clinical and laboratory variables. Results: Performance metrics show high discrimination for 28-day and clinically meaningful association with 1-year mortality, outperforming conventional prognostic scores. Conclusion: CanICU supports objective ICU care allocation via user-friendly implementation.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages13

Reading Time33 min

AccessFree to download

Introduction

Patients and Data Sources

Model Development (CanICU)

Validation Strategy and Performance Metrics

Comparison With Existing Prognostic Scores

External Validation Across Datasets

Conclusion and Clinical Application

cancers Article Machine Learning-Based Mortality Prediction Model for Critically Ill Cancer Patients Admitted to the Intensive Care Unit (CanICU) Ryoung-Eun Ko 1,†, Jaehyeong Cho 2,3,†, Min-Kyue Shin 4, Sung Woo Oh 5, Yeonchan Seong 2,3, Jeongseok Jeon 4, Kyeongman Jeon 1,6, Soonmyung Paik 7, Joon Seok Lim 3,8, Sang Joon Shin 3,9, Joong Bae Ahn 9, Jong Hyuck Park 10, Seng Chan You 2,3,* and Han Sang Kim 3,9,11, * Citation: Ko, R.-E.; Cho, J.; Shin, M.-K.; Oh, S.W.; Seong, Y.; Jeon, J.; Jeon, K.; Paik, S.; Lim, J.S.; Shin, S.J.; et al. Machine Learning-Based Mortality Prediction Model for Critically Ill Cancer Patients Admitted to the Intensive Care Unit (CanICU) . Cancers 2023, 15, 569 . [https://doi.org/10.3390/](https://doi.org/10.3390/)[ ](https://doi.org/10.3390/)cancers15030569 Academic Editor: David Wong Received: 9 November 2022 Revised: 30 December 2022 Accepted: 13 January 2023 Published: 17 January 2023 Copyright: © 2023 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license ([https://](https://)[ ](https://)[creativecommons.org/licenses/by/](creativecommons.org/licenses/by/)[ ](creativecommons.org/licenses/by/)[4.0/](4.0/)) . 1 Department of Critical Care Medicine, Samsung Medical Center, Sungkyunkwan University School of Medicine, Seoul 06351, Republic of Korea 2 Department of Biomedical Systems Informatics, Yonsei University College of Medicine, Seoul 03722, Republic of Korea 3 Institute for Innovation in Digital Healthcare (IIDH), Severance Hospital, Seoul 03722, Republic of Korea 4 Yonsei University College of Medicine, Seoul 03722, Republic of Korea 5 KB Kookmin Bank, Seoul 04534, Republic of Korea 6 Division of Pulmonary and Critical Care Medicine, Department of Medicine, Samsung Medical Center, Sungkyunkwan University School of Medicine, Seoul 06351, Republic of Korea 7 Theragen Bio, Seongnam-si 13488, Republic of Korea 8 Department of Radiology, Yonsei University College of Medicine, Seoul 03722, Republic of Korea 9 Yonsei Cancer Center, Division of Medical Oncology, Department of Internal Medicine, Yonsei University College of Medicine, Seoul 03722, Republic of Korea 10 KAKAO Brain, Seongnam-si 13529, Republic of Korea 11 Graduate School of Medical Science, Brain Korea 21 Project, Yonsei University College of Medicine, Seoul 03722, Republic of Korea * Correspondence: [chandryou@yuhs.ac](chandryou@yuhs.ac) (S.C.Y.); [modeerfhs@yuhs.ac](modeerfhs@yuhs.ac) (H.S.K.); Tel.: +82-2-2228-0363 (S.C.Y.);+82-2-2228-0913 (H.S.K.) † These authors contributed equally to this work. Simple Summary: This study describes a new machine-learning-based 28-day mortality prediction model in adult cancer patients admitted to the intensive care unit (ICU) . A total of 6900 patients in three patient cohorts were used for the development, internal validation, and external validation, respectively, leading to the generation of a reliable model with high sensitivity and speciﬁcity. The CanICU uses nine variables that can be easily obtained in a practical ICU, with the potential beneﬁtof critical care and avoiding unnecessary suffering. Furthermore, this is the largest patient cohort for developing a cancer patient-speciﬁc model. CanICU offers improved performance for predicting short-and long-term mortality in critically ill cancer patients admitted to the ICU. CanICU can help physicians determine how to allocate ICU care for patients with cancer according to objective mortality risk. Abstract: Background: Although cancer patients are increasingly admitted to the intensive care unit (ICU) for cancer-or treatment-related complications, improved mortality prediction remains a big challenge. This study describes a new ML-based mortality prediction model for critically ill cancer patients admitted to ICU. Patients and Methods: We developed CanICU, a machine lear

FAQs

What problem does the CanICU model address?

CanICU addresses the difficulty of improving mortality prediction for critically ill cancer patients admitted to the ICU due to cancer- or treatment-related complications.

How is CanICU constructed and what inputs does it use?

CanICU is a machine-learning model built with a random forest algorithm using nine clinical and laboratory variables that can be obtained in routine ICU practice.

How was CanICU evaluated and how does it compare to existing scores?

CanICU was trained and validated using 6900 patients across internal and external cohorts, reporting sensitivity, specificity, and AUROC for 28-day mortality, and showing better performance than models such as APACHE and SOFA.

Aug 5, 2026

Upload by Patrick