Cervical cancer survival prediction by machine learning algorithms: a systematic review

Area of Science:

• Oncology
• Bioinformatics
• Medical Informatics

Background:

• Cervical cancer remains a significant global health challenge, contributing to female mortality.
• Survival prediction is critical in clinical research for managing cervical cancer.
• Machine learning (ML) offers advanced methods for time-to-event analysis in oncology.

Approach:

• Systematic literature review of PubMed, Scopus, and Web of Science databases up to October 2022.
• Included studies focused on ML algorithms for predicting cervical cancer survival.
• Extracted data included model types, dataset characteristics, and performance metrics like AUC.

Key Points:

• 13 articles published from 2018 onwards were analyzed.
• Random forest, logistic regression, and support vector machines were common ML models.
• Identified 15 variables crucial for predicting cervical cancer survival, with AUCs varying by survival endpoint.

Conclusions:

• Integrating diverse data with ML enhances cervical cancer survival prediction.
• ML interpretability, explainability, and handling imbalanced data are key challenges.
• Further studies are needed to establish ML survival prediction as a clinical standard.