A Novel Integrated Machine Learning Model for Preoperative Evaluation and Postoperative Prognosis-prediction in Cervical Cancer Based on Clinical-pathological Parameters and MR Radiomics (Preprint)

• DOI: 10.2196/preprints.69057
• Publication Date: 2024-11-26T17:57:37Z
• Authors: Shuqi Li; Chenyan Guo; Yufei Fang; Junjun Qiu; He Zhang; Lei Ling; Jie Xu; Xinwei Peng; Chuchu Jiang; Jue Wang; Keqin Hua
• Keywords: Parametrial

<sec> <title>BACKGROUND</title> Machine learning (ML) has been gradually applied to cervical cancer research, but rarely combines both clinical parameters and image data. Meanwhile, more robust accurate preoperative assessment of parametrial invasion lymph node metastasis, as well postoperative prognosis prediction are also in urgent need. </sec> <title>OBJECTIVE</title> We aimed develop an integrated ML model that integrates clinicopathological MR images includes pre- post-operation evaluation (CC) patients. <title>METHODS</title> Data CC patients from 2014 2022 two tertiary hospitals were retrospectively collected exempt was granted by the Ethics Committee for this purpose. Variables analyzed their predictive value invasion, survival recurrence using 7 models. The performance all models compared AI-assisted contouring system is developed based on optimal machine algorithms. <title>RESULTS</title> This study included 250 women analysis (11 deaths, 24 recurrences): (1) In terms invasions with weighted KNN outperformed other models, especially case sensitivity. (2) An achieved predicting times patients, showing high accuracy balanced (3) assists lesion identification, prediction. <title>CONCLUSIONS</title> integration data through offers superior diagnostic prognostic capabilities, potentially reducing errors enabling tailored, precise treatment strategies.