Robustness of Machine Learning Predictions for Determining Whether Deep Inspiration Breath-Hold Is Required in Breast Cancer Radiation Therapy

Aug 2, 2026

Upload by Riley

Deep inspiration breath-hold (DIBH) reduces mean heart dose (MHD), lowering late cardiac side effects in breast cancer patients receiving radiation therapy (RT). Prior machine learning (ML) work has not systematically tested model performance across multiple MHD thresholds and parameter choices. This study evaluates robustness of ML predictions for DIBH need under varied clinical scenarios using 207 patients, testing cut-offs of 240/270/300 cGy, different input-variable counts, and cross-validation folds. Robustness was defined by strong F2 scores and low instability.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages13

Reading Time33 min

AccessFree to download

Abstract

Background/Objectives

Methods

Results

Conclusions

Introduction

Article Robustness of Machine Learning Predictions for Determining Whether Deep Inspiration Breath-Hold Is Required in Breast Cancer Radiation Therapy Wlla E. Al-Hammad 1,2,†, Masahiro Kuroda 3, *,†, Ghaida Al Jamal 2, Mamiko Fujikura 1, Ryo Kamizaki 3,4, Kazuhiro Kuroda 3,5, Suzuka Yoshida 1, Yoshihide Nakamura 1, Masataka Oita 6, Yoshinori Tanabe 3, Kohei Sugimoto 3, Irfan Sugianto 7, Majd Barham 8, Nouha Tekiki 1, Miki Hisatomi 1 and Junichi Asaumi 1 Received: 24 December 2024 Revised: 31 January 2025 Accepted: 6 March 2025 Published: 10 March 2025 Citation: Al-Hammad, W.E.; Kuroda, M.; Al Jamal, G.; Fujikura, M.; Kamizaki, R.; Kuroda, K.; Yoshida, S.; Nakamura, Y.; Oita, M.; Tanabe, Y.; et al. Robustness of Machine Learning Predictions for Determining Whether Deep Inspiration Breath-Hold Is Required in Breast Cancer Radiation Therapy. Diagnostics 2025, 15, 668 . [https://doi.org/10.3390/](https://doi.org/10.3390/)[ ](https://doi.org/10.3390/)diagnostics15060668 Copyright: © 2025 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://creativecommons.org/](https://creativecommons.org/)[ ](https://creativecommons.org/)[licenses/by/4.0/](licenses/by/4.0/)) . 1 Department of Oral and Maxillofacial Radiology, Graduate School of Medicine, Dentistry and Pharmaceutical Sciences, Okayama University, Okayama 700-8558, Japan; [wealhammad@just.edu.jo](wealhammad@just.edu.jo) (W.E.A.-H.) 2 Department of Oral Medicine and Oral Surgery, Faculty of Dentistry, Jordan University of Science and Technology, Irbid 22110, Jordan 3 Radiological Technology, Graduate School of Health Sciences, Okayama University, Okayama 700-8558, Japan 4 Department of Radiology, Matsuyama Red Cross Hospital, Matsuyama 790-8524, Japan 5 Department of Health and Welfare Science, Graduate School of Health and Welfare Science, Okayama Prefectural University, Okayama 719-1197, Japan 6 Graduate School of Interdisciplinary Sciences and Engineering in Health Systems, Okayama University, Okayama 770-8558, Japan 7 Department of Oral Radiology, Faculty of Dentistry, Hasanuddin University, Sulawesi 90245, Indonesia 8 Department of Dentistry and Dental Surgery, College of Medicine and Health Sciences, An-Najah National University, Nablus 44839, Palestine * Correspondence: [kurodamd@cc.okayama-u.ac.jp](kurodamd@cc.okayama-u.ac.jp) † These authors contributed equally to this work. Abstract: Background/Objectives: Deep inspiration breath-hold (DIBH) is a commonly used technique to reduce the mean heart dose (MHD), which is critical for minimizing late cardiac side effects in breast cancer patients undergoing radiation therapy (RT) . Although previous studies have explored the potential of machine learning (ML) to predict which patients might benefit from DIBH, none have rigorously assessed ML model performance across various MHD thresholds and parameter settings. This study aims to evaluate the robustness of ML models in predicting the need for DIBH across different clinical scenarios. Methods: Using data from 207 breast cancer patients treated with RT, we developed and tested ML models at three MHD cut-off values (240, 270, and 300 cGy), considering variations in the number of independent variables (three vs. six) and folds in the crossvalidation (three, four, and five) . Robustness was defined as achieving high F2 scores and low instability in predictive performance. Results: Our findings indicate that the decision tree (DT) model demonstrated consistently high robustness at 240 and 270 cGy, while the random forest model performed optimally at 300 cGy. At 240 cGy, a threshold critical to minimize late cardiac risks, the DT model exhibited stable predictive power, reducing the risk of overestimating DIBH necessity. Conclusions: These results suggest that the DT model, particularly at lower MHD threshol

FAQs

What clinical goal does deep inspiration breath-hold (DIBH) serve in breast cancer RT?

DIBH is used to reduce mean heart dose (MHD), which helps minimize late cardiac side effects in patients receiving breast radiation therapy.

How was robustness of machine learning predictions defined in the study?

Robustness was defined as achieving high F2 scores while maintaining low instability in predictive performance across modeling settings.

Which ML models performed best at different MHD thresholds?

The decision tree model showed consistently high robustness at 240 and 270 cGy, while the random forest model performed best at 300 cGy.

Aug 2, 2026

Upload by Riley