Introduction Machine learning and radiomics (ML/RM) are gaining interest in ovarian cancer (OC) but only few studies have used these methods to predict treatment response. The objective of this study was to review the literature on application of ML/RM in OC and focus on studies describing algorithms to predict treatment response and survival.

Methods A systematic review of published literature from January 1985 to December 2023 on the use of ML/RM in OC was conducted using electronic library databases. P-values were generated using the Pearson’s Chi-squared(x²) test to compare performance of ML/RM models with traditional statistics.

Results Of the 5576 screened articles, 225 studies were included. Between 2021 and 2023, 49 studies were published, highlighting the rapidly growing interest in ML/RM. Median MINORS quality score was similar between studies published from 1985-2021 and 2021-2023(both 8). Neural Network (22.6%) and LASSO (15.3%) were the most common ML/RM algorithms in OC. Among these studies, 10 reported treatment response prediction using radiomics. A total of 3776 patients were analyzed. The most common algorithm was Neural Networks (3/10). Radiomic analysis was used to predict response to neoadjuvant chemotherapy in 6 studies and optimal or complete cytoreduction in 4 studies with a median AUC of 0.77 (range 0.72-0.93) and 0.82 (range 0.77-0.89), respectively. Median model accuracy reported in 5/10 studies was 73% (range 66%-92%).

Conclusion/Implications The use of ML/RM algorithms is becoming a more frequent method to predict response to treatment in OC. These models should be validated in a prospective multicenter trial prior to integration into clinical use.