Developing a Novel Machine Learning-Based Classification Scheme for Predicting SPCs in Breast Cancer Survivors

Chang, Chi-Chang and Chen, Ssu-Han (2019) Developing a Novel Machine Learning-Based Classification Scheme for Predicting SPCs in Breast Cancer Survivors. Frontiers in Genetics, 10. ISSN 1664-8021

[thumbnail of pubmed-zip/versions/2/package-entries/fgene-10-00848.pdf] Text
pubmed-zip/versions/2/package-entries/fgene-10-00848.pdf - Published Version

Download (1MB)

Abstract

Due to the high effectiveness of cancer screening and therapies, the diagnosis of second primary cancers (SPCs) has increased in women with breast cancer. The present study was conducted to develop a novel machine learning–based classification scheme for predicting the risk factors of SPCs in breast cancer survivors. The proposed scheme was based on the XGBoost classifier with the following four comparable strategies: transformation, resampling, clustering, and ensemble learning, to improve the training balanced accuracy. Results suggested that the best prediction accuracy for an empirical case is the XGBoost associated with the strategies of resampling and clustering. The experimental results showed that age, sequence of radiotherapy and surgery, surgical margins of the primary site, human epidermal growth factor, high-dose clinical target volume, and estrogen receptors are relatively more important risk factors associated with SPCs in patients with breast cancer. These risk factors should be monitored for the early detection of breast cancer. In conclusion, the proposed scheme can support the important influence of personality and clinical symptom representations in all phases of the primary treatment trajectory. Our results further suggested that adaptive machine learning techniques require the incorporation of significant variables for optimal predictions.

Item Type: Article
Subjects: Apsci Archives > Medical Science
Depositing User: Unnamed user with email support@apsciarchives.com
Date Deposited: 10 Feb 2023 09:45
Last Modified: 08 Mar 2024 04:29
URI: http://eprints.go2submission.com/id/eprint/366

Actions (login required)

View Item
View Item