Key points are not available for this paper at this time.
PURPOSE OF REVIEW: Advanced gynecologic cancers are treated broadly, with variable outcomes, even when selected by traditional precision oncology biomarkers. Alternative artificial intelligence (AI)-based biomarkers may improve patient selection, cost efficiency and reduce turnaround times. This review summarizes recent evidence regarding application of AI models for prognostic and predictive biomarker development in gynecological malignancies. RECENT FINDINGS: AI-based biomarker discovery has focused on harnessing computational pathology and radiomics machine and deep learning models to infer molecular subtypes, BRCA 1/2 and homologous recombination status, as well as platinum and maintenance therapy sensitivity. However, small samples sizes, modest discriminative power, lack of explainability and of prospective validation are significant limitations. SUMMARY: There is retrospective evidence that AI models potentially constitute useful approaches for precision oncology biomarker detection in advanced gynecological cancers, conditional to prospective validation.
Alonso et al. (Fri,) studied this question.