Tensor fusion improves cancer risk prediction from multiple data types
Tensor-based Fusion of Multi-modal Data for Predicting Cancer Risk Levels
Machine Learning
Summary
Cancer doctors use lots of different tests that create complex data about a patient’s disease. The authors designed a computer method that combines different types of information, like genetic data and clinical details, into a single analysis. This helps find hidden patterns that better predict how risky a patient's cancer might be. Their method works better than using just one type of data alone. They tested it on data from over 11,000 cancer patients across many cancer types.
What this means in practice
- •For hospital data teams: Improve patient risk stratification by integrating diverse clinical and genomic cancer data to generate more accurate risk categories for personalized treatment decisions.
- •For bioinformatics engineers: Develop data processing pipelines that fuse multi-modal biomedical data using tensor methods to enhance feature extraction and clustering for disease outcome prediction.
Authors
Ali Anaissi, Yuanzhe Jia, Qishun Peng, Weidong Huang, Ali Braytee, Junaid Akram
Abstract
Cancer diagnosis and treatment increasingly rely on multi-modal diagnostic technologies that generate high-dimensional datasets. Although these datasets are rich in pathological information, they pose considerable challenges in interpretation, dimensionality reduction, and clinical applicability. Conventional machine learning approaches often fail to extract meaningful cross-modal features, thereby limiting predictive accuracy in patient risk stratification. To address these limitations, this paper proposes an unsupervised machine learning framework that integrates feature extraction, tensor decomposition, and clustering to improve cancer risk prediction. The framework employs tensor decomposition to fuse heterogeneous genomic and clinical data, enabling the extraction of latent, biologically meaningful features that outperform unimodal feature sets. Multiple feature extraction strategies are investigated, and the resulting features are subsequently clustered using a range of clustering algorithms. Comparative evaluations based on internal and external metrics are then used to identify the most effective configurations. The framework is validated on LinkedOmics data comprising 11,158 patients across 32 cancer types. For patient risk stratification, clinically relevant attributes such as tumor purity, pathological stage, and survival outcomes serve as ground-truth indicators. Experimental results demonstrate the superiority of tSVD+GMM, which consistently outperforms other configurations across clinical benchmarks.