Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2023 Jan 18:2023:7931321.
doi: 10.1155/2023/7931321. eCollection 2023.

Multiview Deep Forest for Overall Survival Prediction in Cancer

Qiucen Li  1 , Zedong Du  2 , Zhikui Chen  1 , Xiaodi Huang  3 , Qiu Li  4
Affiliations

Multiview Deep Forest for Overall Survival Prediction in Cancer

Qiucen Li et al. Comput Math Methods Med. .

Abstract

Overall survival (OS) in cancer is crucial for cancer treatment. Many machine learning methods have been applied to predict OS, but there are still the challenges of dealing with multiview data and overfitting. To overcome these problems, we propose a multiview deep forest (MVDF) in this paper. MVDF can learn the features of each view and fuse them with integrated learning and multiple kernel learning. Then, a gradient boost forest based on the information bottleneck theory is proposed to reduce redundant information and avoid overfitting. In addition, a pruning strategy for a cascaded forest is used to limit the impact of outlier data. Comprehensive experiments have been carried out on a data set from West China Hospital of Sichuan University and two public data sets. Results have demonstrated that our method outperforms the compared methods in predicting overall survival.

PubMed Disclaimer

Conflict of interest statement

The authors declare that they have no conflicts of interest.

Figures

Figure 1
Figure 1
The overall procedure of multiview deep forest (MVDF). The total number of levels of cascaded forests is determined automatically.
Figure 2
Figure 2
Multiview feature extraction. (a) Views are formed according to their relationship between original data. (b) Knowledge learned in random forests is used to select important indicators.
Figure 3
Figure 3
Pruning cascade forest. According to the classification results and Gini value of leaf nodes, each tree in the forest was evaluated and formed a probability vector. Pruning the forest according to the vector.
Figure 4
Figure 4
Performance under precision and recall comparisons of different approaches on three classes.
Figure 5
Figure 5
Sensitivity analysis of β in the function of Equation (3).
Figure 6
Figure 6
A decision tree in MVDF and the important indicators selected by forests and doctors.
Figure 7
Figure 7
Sensitivity analysis of the parameters m and h in MVDF.
See this image and copyright information in PMC

References

    1. N C Institute. Understanding Cancer Prognosis, Website . February 2022, https://www.cancer.gov/about-cancer/diagnosis-staging/prognosis/
    1. Smith J. C., Sheltzer J. M. Systematic identification of mutations and copy number alterations associated with cancer patient prognosis. eLife . 2018;7, article e39217 doi: 10.7554/eLife.39217. - DOI - PMC - PubMed
    1. Zhao Z., Bai L., Zhang Y., Han J. Classification via semi-supervised multirandom subspace sparse representation. Signal, Image and Video Processing . 2019;13(7):1387–1394. doi: 10.1007/s11760-019-01467-8. - DOI
    1. Hatherley J. J. Limits of trust in medical AI. Journal of Medical Ethics . 2020;46(7):478–481. doi: 10.1136/medethics-2019-105935. - DOI - PubMed
    1. Che S., Peng H., Sun L., Chen Y., He L. Federated multi-view learning for private medical data integration and analysis. 2021. https://arxiv.org/abs/2105.01603.