Hira, MT, Razzaque, MA and Sarker, M (2024) Ovarian cancer data analysis using deep learning: A systematic review. Engineering Applications of Artificial Intelligence, 138. ISSN 0952-1976

Preview

Text Ovarian cancer data analysis using deep learning A systematic review.pdf - Published Version Available under License Creative Commons Attribution. Download (2MB) | Preview

Abstract

Technological advancement and the adoption of digital technologies in cancer care and research have generated big data. These diverse and multimodal data contain valuable high-density information on various cancer subdomains, including early detection and accurate diagnosis. By extracting this information, machine learning or deep learning (ML or DL)-based autonomous data analysis tools can help clinicians and cancer researchers discover patterns and relationships from complex datasets. Many DL-based analyses on ovarian cancer (OC) data have recently been published. These analyses are highly diverse in various aspects or features of cancer (e.g., subdomain(s) and cancer type they address) and data analysis features (e.g., data modality, analysis method). However, a comprehensive understanding of these analyses in terms of these features and Artificial Intelligence Assurance (AIA) is currently lacking. This systematic review aims to fill this gap by examining the existing literature and identifying important aspects of OC data analysis using DL, explicitly focusing on key features and AI assurance perspectives. We used the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) framework for detailed searches across three journal databases and included only peer-reviewed studies published from 2015 to 2023. We identified and reviewed 96 DL-based OC studies and found five important insights on DL-based OC data analysis. First, most studies 71% (68 of 96) focused on detection and diagnosis, while no study addressed the prediction and prevention of OC. Second, the analyses were predominantly based on samples from a non-diverse population (75% (72/96 studies)), limited to a geographic location or country. Third, only a small proportion of the studies (only 33% (32/96)) performed integrated analyses, most of which used homogeneous data (clinical or omics). Fourth, only 8.3% (8/96) of the studies validated their models using external and diverse datasets, highlighting the need for enhanced model validation, and finally, the inclusion of AIA in cancer data analysis is in a very early stage; only 5.2% (5/96) explicitly addressed AIA through explainability. We also highlight critical areas that require attention in DL-based cancer data analysis, especially OC data analysis. Future research should address identified gaps, including exploring diverse and heterogeneous integrated data-driven analyses, validating models using external datasets from different demographic populations, and focusing on AI assurance through all aspects, including explainability and safety.

Item Type:	Article
Uncontrolled Keywords:	Deep learning; Machine learning; Ovarian cancer; Precision oncology; Multi-omics; Integrated analysis; Hybrid model; External validation; Detection and diagnosis; Prognosis; 46 Information and Computing Sciences; 40 Engineering; Bioengineering; Data Science; Cancer; Ovarian Cancer; Machine Learning and Artificial Intelligence; Rare Diseases; Networking and Information Technology R&D (NITRD); Prevention; Women's Health; Cancer; 08 Information and Computing Sciences; 09 Engineering; Artificial Intelligence & Image Processing; 40 Engineering; 46 Information and computing sciences
Subjects:	Q Science > QH Natural history > QH301 Biology R Medicine > R Medicine (General)
Divisions:	Pharmacy and Biomolecular Sciences
Publisher:	Elsevier
SWORD Depositor:	A Symplectic
Date Deposited:	09 Apr 2025 14:08
Last Modified:	09 Apr 2025 14:15
DOI or ID number:	10.1016/j.engappai.2024.109250
URI:	https://researchonline.ljmu.ac.uk/id/eprint/26142

View Item