The fusion of data features from different modes, such as pathology images and sequence data, has the potential to predict the overall survival (OS) of patients with cervical cancer. This study aims to develop a novel prediction model for overall survival (OS) that incorporates pathology images, clinical data and molecular data. The model underwent training using comprehensive cervical cancer data from The Cancer Genome Atlas (TCGA), which include 119 patients. To independently validate the model, we used a manually collected dataset from Peking Union Medical College Hospital (PUMCH), comprising 53 patients with cervical cancer. LASSO Cox regression analysis was applied to identify relevant features associated with overall survival (OS), resulting in the identification of 484 genes, including RGR, DBN1 and CALCR, as well as numerous image features. Building upon these findings, a multimodal deep learning model was developed to effectively classify the overall survival (OS) of patients with cervical cancer into two categories: short term (ST: ≤ 3 years) and long term (LT: > 3 years) based on the integration of pathology images and clinical features. The developed model achieved reasonably good prediction accuracy in the independent testing dataset from PUMCH, with an area under the curve (AUC) value of 0.783. In conclusion, the combination of pathology images with clinical and molecular data enables the creation of accurate and reliable prediction models for cervical cancer.
Shen et al. (Thu,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: