Machine and Deep Learning for the Diagnosis, Prognosis, and Treatment of Cervical Cancer: A Scoping Review.

Abstract

Background/Objectives: Cervical cancer (CC) is the fourth most common cancer among women worldwide. This study explored the use of machine learning (ML) and deep learning (DL) in the prediction, diagnosis, and prognosis of CC. Methods: An electronic search was conducted in the PubMed, IEEE, Web of Science, and Scopus databases from January 2015 to April 2025 using the search terms ML, DL, and uterine cervical neoplasms. A total of 153 studies were selected in this review. A comprehensive summary of the available evidence was compiled. Results: We found that 54.9% of the studies addressed the application of ML and DL in CC for diagnostic purposes, followed by prognosis (22.9%) and an incipient focus on CC treatment (22.2%). The five countries where most ML and DL applications have been generated are China, the United States, India, Republic of Korea, and Japan. Of these studies, 48.4% proposed a DL-based approach, and the most frequent input data used to train the models on CC were images. Conclusions: Although there are results indicating a promising application of these artificial intelligence approaches in oncology clinical practice, further evidence of their validity and reproducibility is required for their use in early detection, prognosis, and therapeutic management of CC.