Intelligent Monitoring and Analysis Algorithm for C17ORF49 Gene Expression in ERα-Positive Breast Cancer
Summary
No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.
Summarise with VIRAT Free to try with your Science Catalogs account.Abstract
To address the dimensional limitations of traditional single-modal analysis in clinical decision-making for ERα-positive breast cancer, a multimodal feature fusion attention algorithm (MFFA) was proposed. This algorithm employs a four-layer progressive architecture. The input layer constructs a multidimensional system by selecting RNA-seq gene data (including C17ORF49), clinical pathology data, and pathology slide imaging features. The feature extraction layer processes gene and time-series data using an improved CNN and bidirectional LSTM, respectively. The attention fusion layer utilizes dynamic weighting to enhance key features. The output layer performs dual-task risk stratification and gene expression prediction in parallel. The experiment utilized dual-source data from the TCGA database (1200 cases) and a local hospital (300 cases). After preprocessing with KNN interpolation and 3σ outlier removal, the dataset was partitioned into a 7:1.5:1.5 split. Using NVIDIA A100 hardware, MFFA was compared with SVM, XGBoost, and a standard attention model. The results showed that MFFA achieved an Accuracy of 94.2% and an Area Under the Curve (AUC) of 0.973 for classification tasks, representing improvements of 7.7% and 8.1% respectively over XGBoost. The RMSE for regression tasks was 0.086, a 43.4% reduction over the standard attention model. In a noise interference experiment, MFFA achieved an Accuracy of 89.5% even with 30% noise, and the AUC variation across datasets was only 1.5%, demonstrating its robustness and generalizability, providing support for precision diagnosis and treatment of breast cancer.
Fields of Study & Subjects
Breast Cancer Treatment Studies
Gene expression and cancer classification
AI in cancer detection
Publication details
- Journal / source
- IEEE International Conference on Power, Electronics and Green Energy (ICPEGE)
- Pages
- 771–776
- Publication date
- 28 January 2026
- Language
- EN