Application of artificial intelligence in the correlation between HPV subtypes and cytological patterns: an integrative review

Infection by the human papillomavirus (HPV) is the main cause of cervical cancer and cervical intraepithelial neoplasia (CIN) among women worldwide, and the third leading cause of death in Brazil. The HPV subtypes that infect the genital tract are transmitted through sexual contact and infect various sites. Regarding cervical cancer, several computational scoring systems aided by AI have been proposed in recent years with the aim of optimizing the accuracy of cervical neoplasm diagnosis. This study aimed to analyze, through an Integrative Literature Review (ILR), the correlation between HPV subtypes and cytological patterns. For this purpose, articles published between 2020 and 2025 were selected from the databases BVS, PubMed, Google Scholar, Capes, Ebsco, SciELO, and ResearchRabbit. The descriptors used were: “Human Papillomavirus,” “Papanicolaou Test,” “Machine Learning,” “Cytopathology,” “AI,” “HPV subtypes,” “Colposcopy,” and others in Portuguese and English. The results indicated that methods such as CerviCARE AI, AI-LBC, CAIADS, ResNet10, and TruScreen have a high capacity to recognize cervical lesions. Some models demonstrate diagnostic potential superior to experienced colposcopists, positioning Artificial Intelligence as a viable and promising tool in cervical cancer screening. It is concluded that the research contributes to the maturation of the field of bioinformatics applied to gynecologic oncology, showing that the integration of cytological and molecular data can be effectively interpreted by machine learning models. In practice, the results highlight that these tools can be incorporated into population screening programs, making them more optimized.