Back to Search
Start Over
Cancer prognosis with machine learning-based modified meta-heuristics and weighted gradient boosting algorithm.
- Source :
- Computer Methods in Biomechanics & Biomedical Engineering: Imaging & Visualisation; Dec2023, Vol. 11 Issue 6, p2209-2225, 17p
- Publication Year :
- 2023
-
Abstract
- Cancer is typically a disease that is instigated when cells partition uncontrollably, thereby spread to surrounding tissues. Among different kinds of cancer, breast cancer, cervical and lung cancer have found to experience less screening rate. Hence, detecting such kinds of cancer has become crucial to save individual's life. With the ability of ML (Machine Learning) for solving complicated tasks in dynamic configurations have contributed to its application in prognosticating cancer with diverse meta-heuristic algorithms. Though conventional researches have attempted to perform this, feature selection process has been inefficient that negatively impacted the accuracy rate. To resolve such consequences, the study considers Wisconsin breast cancer dataset, lung cancer dataset and cervical cancer dataset from UCI ML repository and proposes Modified QPSO-LDA (Quantum Particle Swarm Optimization-Linear Discriminant Analysis) to select suitable and relevant features. Following this, WGB (Weighted Gradient Boosting) is proposed for classification that works based on weighted column subsampling. To validate the performance of this system in prognosticating breast cancer, cervical cancer and lung cancer, analysis is undertaken that confirm its effectiveness with 0.9912 as accuracy for breast cancer dataset, 0.99 as accuracy for cervical cancer dataset and 0.97 as accuracy for lung cancer dataset. [ABSTRACT FROM AUTHOR]
Details
- Language :
- English
- ISSN :
- 21681163
- Volume :
- 11
- Issue :
- 6
- Database :
- Complementary Index
- Journal :
- Computer Methods in Biomechanics & Biomedical Engineering: Imaging & Visualisation
- Publication Type :
- Academic Journal
- Accession number :
- 174632734
- Full Text :
- https://doi.org/10.1080/21681163.2023.2219772