S408
Clinical - Gynaecological
ESTRO 2026
Digital Poster 695
Cancer and Gynecologic Malignancies (MITO) group conducted a systematic literature review to map the current landscape of AI- based predictive modelling in patients with gynecological malignancies treated with radiotherapy (RT). Material/Methods:
Clinical challenges in the adaptation of AI predictive models in radiation oncology for gynecological cancer: a systematic review by the AI MITO Group Donato Pezzulla 1 , Amelia Barcellini 2,3 , Savino Cilla 4 , Michele Aquilano 5 , Paolo Bonome 1 , Alexandra Charalampopoulou 6 , Gaia Giannone 7 , Valentina Lombardo 8 , Federico Mastroleo 9 , Maurizio Polano 10 , Ivano Raimondo 11,12 , Sandro Pignata 13 , Gabriella Macchia 1 1 Radiation Oncology Unit, Responsible Research Hospital, Campobasso, Italy. 2 Radiation oncology Unit, Clinical Department, CNAO National Center of Oncological Hadrontherapy, Pavia, Italy. 3 Department of Internal Medicine and Therapeutics, University of Pavia, Pavia, Italy. 4 Medical Physics Unit, Responsible Research Hospital, Campobasso, Italy. 5 Cyberknife Center, Istituto Fiorentino Di Cura Ed Assistenza (IFCA), University of Florence, Florence, Italy. 6 Radiobiologi Unit, Research and Development Department, CNAO National Center for Oncological Hadrontherapy, Pavia, Italy. 7 Department of Surgery and Cancer, Ovarian Cancer Action Research Centre, Imperial College London, London, United Kingdom. 8 Medicine and Surgery, Kore University, Enna, Italy. 9 Division of Radiation Oncology, European Institute of Oncology IRCCS, Milan, Italy; Department of Oncology and Hemato-oncology, Milan, Italy. 10 Experimental and Clinical Pharmacology Unit, National Cancer Institute C.R.O.-IRCCS Aviano, Aviano, Italy. 11 School in Biomedical Sciences, University of Sassari, Sassari, Italy. 12 Department of Gynecology, Mater Olbia Hospital, Olbia, Italy. 13 Uro-Gynecological Department, National Cancer Institute of Naples Fondazione G Pascale IRCCS, Naples, Italy Purpose/Objective: The Radiation Oncology-AI Subcommittee of the Multicenter Italian Trials in Ovarian
Relevant studies were systematically retrieved from electronic databases
following PRISMA Guidelines. The Critical Appraisal and Data Extraction for Systematic Reviews of Prediction Modelling Studies (CHARMS) checklist was applied to evaluate methodological quality, while the Prediction model Risk Of Bias Assessment Tool (PROBAST) was applied to estimate risk of bias and clinical applicability. Results: The search yielded 1402 records, of which 1025 unique studies were screened after duplicate removal. Following eligibility assessment, 31 studies (1.1%) met the inclusion criteria (Figure 1).
Among these, 27 (87.0%) focused on cervical cancer, 2 (6.5%) on uterine tumors, and 2 (6.5%) on mixed gynecologic populations. Most predictive models were developed to assess clinical outcomes (11/31; 35.5%), treatment-related adverse events (9/31; 29.0%), or treatment response (9/31; 29.0%). The number of predictors included in the models ranged from 3 to 114, incorporating clinical, dosimetric, and radiomic features,
Made with FlippingBook - Share PDF online