Predicción diagnóstica de la endometriosis en mujeres en edad reproductiva mediante inteligencia artificial: una revisión sistemática

Abstract

Title: Diagnostic Prediction of Endometriosis in Women of Reproductive Age Using Artificial Intelligence: A Systematic Review Objective: To evaluate the methods and outcomes of using artificial intelligence for the diagnostic prediction of endometriosis in women of reproductive age, to determine its accuracy and clinical feasibility. Methodology: A systematic review was conducted based on the scientific literature available in the following databases: PubMed, Wiley Online Library, PLOS ONE, ScienceDirect, arXiv, Sage Journals, and SpringerLink. A total of 646 studies were identified and screened according to the PRISMA guidelines. Of these, 14 scientific articles relevant to the research topic and published within the last five years met all criteria and were included. Results: Of the 646 articles retrieved, 2.16% (n=14) met the inclusion criteria. All included studies (100%) reported significant improvements in diagnostic accuracy using artificial intelligence: 64.3% (n=9) used image-based models, 28.6% (n=4) analyzed biomarkers and clinical data, and 21.4% (n=3) evaluated mobile apps and self-reports. Furthermore, 71.4% (n=10) reported an area under the curve (AUC) greater than 0.87, and 42.8% (n=6) achieved a sensitivity greater than 90%, highlighting the effectiveness of these methods and supporting their clinical feasibility.

Conclusions: Artificial intelligence improves endometriosis diagnosis by offering noninvasive, personalized, and highly accurate approaches. Although it still needs validation and regulation, integrating it into clinical practice can support more efficient, patient-centered care.