Scholarlink

Background <p>Breast cancer is a type of disease that has caused significant mortality and morbidity worldwide, making diagnosis and prognosis crucial for managing the treatment process of patients. In recent years, artificial intelligence and machine learning have brought significant advancements in the field of medicine; among the advantages of these technologies in healthcare are the faster prediction of diseases and the reduction of their complications.</p> Methods <p>This applied study was conducted in 2024 in Iran. Two hundred and ten female patients with left breast cancer were selected for the study. Multi-label classification algorithms were utilized to predict the radiation exposure to the heart and lungs, and their performance was evaluated using Accuracy, Precision, Recall, F1-Score, Hamming Loss, and Area Under the ROC Curve.</p> Results <p>Based on the implementation results of the algorithms, the KNN algorithm achieved the best performance among the other algorithms, with an accuracy of 44.1%, precision of 73.9%, recall of 70.9%, F1 score of 71.6%, AUC of 72.2%, and Hamming loss of 27.00%.</p> Conclusions <p>The findings of this study indicate that anatomical features significantly affect the radiation doses received by the heart and lungs of breast cancer patients; however, the impact of anatomical variables varies among each other. In terms of influence, the width and thickness of the breast are two highly impactful variables on the amount of radiation exposure during radiation therapy. Machine learning algorithms showed promising performance in predicting the radiation exposure to the heart and lungs, with the KNN algorithm yielding the best results among them.</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

Evaluation of Machine Learning Algorithms in Predicting Cardiac and Pulmonary Doses for Breast Cancer Patients Undergoing Radiation Therapy: A Multi-Label Classification Approach with Feature Selection

  • 期刊论文

  • 发表日期：

    2026年2月25日

查看全文
  • Pooria Afsharifard,
  • Mahshid Abbasi,
  • Mohammad Bagher Dowlatshahi,
  • Abbas Rezaeian,
  • Bagher Goldoost,
  • Morteza Amraei

摘要

Background

Breast cancer is a type of disease that has caused significant mortality and morbidity worldwide, making diagnosis and prognosis crucial for managing the treatment process of patients. In recent years, artificial intelligence and machine learning have brought significant advancements in the field of medicine; among the advantages of these technologies in healthcare are the faster prediction of diseases and the reduction of their complications.

Methods

This applied study was conducted in 2024 in Iran. Two hundred and ten female patients with left breast cancer were selected for the study. Multi-label classification algorithms were utilized to predict the radiation exposure to the heart and lungs, and their performance was evaluated using Accuracy, Precision, Recall, F1-Score, Hamming Loss, and Area Under the ROC Curve.

Results

Based on the implementation results of the algorithms, the KNN algorithm achieved the best performance among the other algorithms, with an accuracy of 44.1%, precision of 73.9%, recall of 70.9%, F1 score of 71.6%, AUC of 72.2%, and Hamming loss of 27.00%.

Conclusions

The findings of this study indicate that anatomical features significantly affect the radiation doses received by the heart and lungs of breast cancer patients; however, the impact of anatomical variables varies among each other. In terms of influence, the width and thickness of the breast are two highly impactful variables on the amount of radiation exposure during radiation therapy. Machine learning algorithms showed promising performance in predicting the radiation exposure to the heart and lungs, with the KNN algorithm yielding the best results among them.