Tue, Nov 3, 2026
[Archive]
Volume 18, Issue 3 (September-2026 2026)
Iranian Journal of Blood and Cancer 2026, 18(3): 64-76 |
Back to browse issues page
Download citation:
BibTeX | RIS | EndNote | Medlars | ProCite | Reference Manager | RefWorks
Send citation to:
BibTeX | RIS | EndNote | Medlars | ProCite | Reference Manager | RefWorks
Send citation to:
Akbari A, Rahimi M, Asadi F, Akbari M E. Explainable Deep Learning for Ovarian Cancer Survival Prediction: A Retrospective Analysis of Iranian National Data. , Iranian Journal of Blood and Cancer. 2026; 18 (3) :64-76
URL: http://ijbc.ir/article-1-1922-en.html
URL: http://ijbc.ir/article-1-1922-en.html
1- Cancer Research Center, Shahid Beheshti University of Medical Sciences, Tehran, Iran.
2- Health and Biomedical Informatics Research Center, Urmia University of Medical Sciences, Urmia, Iran.
3- Department of Health Information Technology and Management, School of Allied Medical Sciences, Shahid Beheshti University of Medical Sciences, Tehran, Iran. ,asadifar@sbmu.ac.ir
2- Health and Biomedical Informatics Research Center, Urmia University of Medical Sciences, Urmia, Iran.
3- Department of Health Information Technology and Management, School of Allied Medical Sciences, Shahid Beheshti University of Medical Sciences, Tehran, Iran. ,
Abstract: (11 Views)
Background: Ovarian cancer is one of the deadliest gynecological cancers, creating an urgent need for reliable prognostic tools to guide treatment decisions. With the growing use of deep learning in healthcare, this study aimed to predict ovarian cancer survival outcomes even when working with incomplete or low-quality clinical data.
Methods: We performed a retrospective analysis of 2,675 ovarian cancer cases with complete survival data from the Iranian National Cancer Registry (2008–2015). Survival probabilities were estimated using Kaplan–Meier analysis, and eight deep learning models, covering discrete- and continuous-time approaches, were compared using the concordance index (C-index). Model interpretability analyses were also conducted to identify key prognostic factors.
Results: Five- and ten-year survival rates were 54% and 42%, respectively. The discrete-time Logistic Hazard model achieved the best predictive performance (C-index = 0.69). Age at diagnosis and histological subtype were significant survival predictors (p < 0.05), and patients with epithelial type II tumors had poorer outcomes despite their higher prevalence.
Conclusion: Deep learning shows strong potential for survival prediction in ovarian cancer, even with imperfect data. However, improving data completeness and model transparency is crucial before these tools can be applied in clinical settings.
Methods: We performed a retrospective analysis of 2,675 ovarian cancer cases with complete survival data from the Iranian National Cancer Registry (2008–2015). Survival probabilities were estimated using Kaplan–Meier analysis, and eight deep learning models, covering discrete- and continuous-time approaches, were compared using the concordance index (C-index). Model interpretability analyses were also conducted to identify key prognostic factors.
Results: Five- and ten-year survival rates were 54% and 42%, respectively. The discrete-time Logistic Hazard model achieved the best predictive performance (C-index = 0.69). Age at diagnosis and histological subtype were significant survival predictors (p < 0.05), and patients with epithelial type II tumors had poorer outcomes despite their higher prevalence.
Conclusion: Deep learning shows strong potential for survival prediction in ovarian cancer, even with imperfect data. However, improving data completeness and model transparency is crucial before these tools can be applied in clinical settings.
: Original Article |
Subject:
AI in Medicine
Received: 2026/07/15 | Accepted: 2026/09/20 | Published: 2026/09/30
Received: 2026/07/15 | Accepted: 2026/09/20 | Published: 2026/09/30
Send email to the article author
| Rights and permissions | |
 |
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License. |
- Iranian Journal of Blood and Cancer
- Journal Tel: +989369936116
- Publisher Tel: +982122275917
- Mail: Info@ijbc.ir
