Abstract
The possible contribution of ultrasonography to the prognosis of cervical cancer is still mostly unknown. Clinicopathological information about patients with cervical cancer from 2019 to 2024 was gathered from the database of the Chongqing Medical University Second Affiliated Hospital. A training group comprising 70% of the dataset and a validation cohort comprising 30% were randomly selected. Univariate and multivariate competing risk models were used to identify independent prognostic markers. A nomogram to forecast the probability of death specific to cancer was created based on these variables. ROC curves, area under the curve (AUC), concordance index (C-index), and calibration curves were used to evaluate the nomogram's accuracy and discriminative power. The training and validation sets were validated independently. 428 individuals with cervical cancer were randomized to be in one of two groups: the training group (n = 296) or the validation group (n = 132). 54.0 months was the average follow-up period (range: 6.0-80.0 months). Advanced age (p = 0.002) and FIGO stages III-IV (p = 0.020) were strongly linked to premature mortality. Moreover, no previous surgery and ultrasound-assessed blood supply and interstitial infiltration were also independently correlated with lesser overall survival. These results led to the development of a nomogram to estimate cancer-specific survival at one, two, and three years. In this work, we created and verified a predictive nomogram for patients with cervical cancer, integrating ultrasonography characteristics into a survival prediction model for the first time.