Article
Mathematics
Ala'a El-Nabawy, Nahla A. Belal, Nashwa El-Bendary
Summary: Automated diagnosis systems aim to reduce diagnosis costs with maintained efficiency. The cascade Deep Forest ensemble model shows competitive classification accuracy with other techniques, particularly for imbalanced training sets, by utilizing cascade ensemble decision trees to learn hyper-representations. The use of gene expression data alone with the cascade Deep Forest classifier achieves comparable accuracy to other techniques with higher computational performance, with times recorded around 5-7 seconds.
Article
Gastroenterology & Hepatology
Eileen Morgan, Isabelle Soerjomataram, Anna T. Gavin, Mark J. Rutherford, Piers Gatenby, Aude Bardot, Jacques Ferlay, Oliver Bucher, Prithwish De, Gerda Engholm, Christopher Jackson, Serena Kozie, Alana Little, Bjorn Moller, Lorraine Shack, Hanna Tervonen, Vicky Thursfield, Sally Vernon, Paul M. Walsh, Ryan R. Woods, Christian Finley, Neil Merrett, Dianne L. O'Connell, John Reynolds, Freddie Bray, Melina Arnold
Summary: The survival of esophageal cancer has significantly improved over the past 20 years in participating countries, especially for adenocarcinoma, younger age groups, and the first year after diagnosis. Australia and Ireland consistently showed higher survival rates. However, further advancements in early detection and treatment are needed for specific groups and patients with squamous cell carcinoma.
Article
Biochemical Research Methods
Yunhe Wang, Zhiqiang Ma, Ka-Chun Wong, Xiangtao Li
Summary: Detection and diagnosis of cancer are crucial for early prevention and treatment. This article introduces a multiobjective PSO-based hybrid algorithm (MOPSOHA) for optimizing feature selection and diagnosis of cancer data, which outperforms other algorithms in various cancer datasets. The effectiveness of MOPSOHA is demonstrated through its novel encoding strategy, mutation operator, and local search method.
IEEE-ACM TRANSACTIONS ON COMPUTATIONAL BIOLOGY AND BIOINFORMATICS
(2021)
Article
Oncology
Pratip Rana, Phuc Thai, Thang Dinh, Preetam Ghosh
Summary: A novel feature selection algorithm DTA was introduced for selecting important, non-redundant, and relevant features from diverse omics data, demonstrating superior performance in binary and multiclass classification problems across different diseases. The algorithm improved disease subtyping accuracy for various cancer types, suggesting its potential for supporting biomarker selection, precision medicine design, and disease sub-type detection in the scientific community.
Article
Computer Science, Artificial Intelligence
Jiande Huang, Ping Chen, Lijuan Lu, Yuhui Deng, Qiang Zou
Summary: The paper proposes a Weighted Cascade Deep Forest framework (WCDForest) that addresses overfitting and characteristic dispersion issues in the deep forest model. The framework uses a multi-grained scanning module and a class vector weighting module to enhance performance, and introduces a feature enhancement module to reduce information loss. Experimental results demonstrate that WCDForest outperforms existing models.
APPLIED INTELLIGENCE
(2023)
Article
Biochemical Research Methods
Jiquan Shen, Jiawei Shi, Junwei Luo, Haixia Zhai, Xiaoyan Liu, Zhengjiang Wu, Chaokun Yan, Huimin Luo
Summary: This paper proposes a deep learning approach, DCGN, which combines a CNN and BiGRU to achieve nonlinear dimensionality reduction and feature learning for eliminating irrelevant factors in gene expression data. Experimental results demonstrate that DCGN outperforms seven other cancer subtype classification methods.
BMC BIOINFORMATICS
(2022)
Article
Biochemical Research Methods
Hai Yang, Rui Chen, Dongdong Li, Zhe Wang
Summary: Subtype-GAN is a deep adversarial learning method that accurately identifies molecular subtypes of tumor samples. It performs outstandingly on benchmark datasets and holds theoretical and practical value in analysis.
Article
Oncology
Anna Plotkin, Ekaterina Olkhov-Mitsel, Sharon Nofech-Mozes, Bojana Djordjevic, Jelena Mirkovic, Madeline Fitzpatrick, Adriana Krizova, Nicole J. Look Hong
Summary: This study evaluated the costs associated with four approaches to classifying endometrial cancer and found that histomolecular classification had the highest cost. Therefore, informed decision-making is needed when implementing molecular classification in clinical practice.
GYNECOLOGIC ONCOLOGY
(2023)
Article
Computer Science, Artificial Intelligence
Bo Yu, Hechang Chen, Yunke Zhang, Lele Cong, Shuchao Pang, Hongren Zhou, Ziye Wang, Xianling Cong
Summary: In this paper, a Data and Knowledge Co-driving (D&K) model is proposed to replicate the process of cancer subtype classification on histopathological slides. The model combines data-driven and knowledge-driven modules to achieve accurate and efficient diagnosis of histopathological subtypes.
KNOWLEDGE-BASED SYSTEMS
(2023)
Article
Biochemical Research Methods
Jiali Pang, Bilin Liang, Ruifeng Ding, Qiujuan Yan, Ruiyao Chen, Jie Xu
Summary: We propose a denoised multi-omics integration framework, including a distribution-based feature denoising algorithm (FSD) for dimension reduction, and an Attention Multi-Omics Integration (AttentionMOI) framework for cancer prognosis prediction and subtype identification. We demonstrate that FSD improves model performance in survival prediction and kidney cancer subtype identification using single omic or multi-omics data in 15 TCGA cancers. Our integration framework AttentionMOI outperforms machine learning models and current multi-omics integration algorithms with high dimensions of features. Furthermore, FSD identifies features associated with cancer prognosis that can be considered as biomarkers.
BRIEFINGS IN BIOINFORMATICS
(2023)
Article
Computer Science, Artificial Intelligence
Hongren Zhou, Hechang Chen, Bo Yu, Shuchao Pang, Xianling Cong, Lele Cong
Summary: This paper proposes an end-to-end weakly supervised learning framework called EWSLF to address challenges in histopathology data analysis. The framework utilizes cluster-based sampling and multi-branch attention mechanism to refine histological features and improve classification accuracy for cancer subtype classification. Experimental results demonstrate the superior and credible results of the proposed model compared to state-of-the-art methods.
EXPERT SYSTEMS WITH APPLICATIONS
(2024)
Article
Engineering, Chemical
Suhuan Bi, Liangliang Mu, Xiuyan Liu
Summary: This study introduces an improved algorithm model for classifying and identifying different drying conditions in the tobacco drying process. The method effectively transforms fuzzy artificial judgment into data-driven identification, showing high precision and promising prospects.
Article
Genetics & Heredity
Can Liu, Yuchen Duan, Qingqing Zhou, Yongkang Wang, Yong Gao, Hongxing Kan, Jili Hu
Summary: In this study, a gastric cancer subtype classification model called RRGCN was developed based on multi-omics fusion data and patient similarity network using residual graph convolutional network (GCN). The results demonstrate that RRGCN outperforms other classification methods with a high accuracy of 0.87 compared to traditional machine learning methods and deep learning models. Overall, RRGCN shows great potential in providing fresh perspectives on disease mechanisms and progression, and can be valuable for a wide range of disorders and clinical diagnosis.
FRONTIERS IN GENETICS
(2023)
Article
Environmental Sciences
Shiqi Zhang, Peihao Peng, Maoyang Bai, Xiao Wang, Lifu Zhang, Jiao Hu, Meilian Wang, Xueman Wang, Juan Wang, Donghui Zhang, Xuejian Sun, Xiaoai Dai
Summary: This study proposes a hierarchy-based classifier combined with environmental variables to quantitatively classify the different vegetation subtypes of Sichuan's mountainous evergreen broad-leaved forests. The study reveals the widespread distribution of evergreen broad-leaved forests in Sichuan, with clear boundaries between the distribution areas of the humid and semi-humid subtypes. The methods used in this study offer an effective approach to vegetation classification in mountainous areas and provide guidance for ecological engineering, ecological protection, and agricultural and livestock development.
Article
Engineering, Biomedical
Lu Zhao, Xiaowei Xu, Runping Hou, Wangyuan Zhao, Hai Zhong, Haohua Teng, Yuchen Han, Xiaolong Fu, Jianqi Sun, Jun Zhao
Summary: This paper proposes a weakly supervised framework for accurate subtype classification of non-small-cell lung cancer through a two-stage structure of ROI localization and subtype classification. Experimental results show an AUC of 0.9602 in ROI localization and 0.9671 in subtype classification. The proposed method demonstrates superiority in NSCLC subtype classification and can be extended to other classification tasks with WSIs.
PHYSICS IN MEDICINE AND BIOLOGY
(2021)