Non-negative matrix factorization model-based construction for molecular clustering and prognostic assessment of head and neck squamous carcinoma.

ABSTRACT

Purpose: We aimed at exploring the efficacy of non-negative matrix factorization (NMF) model-based clustering for prognostic assessment of head and neck squamous carcinoma (HNSCC). Methods: The transcriptome microarray data of HNSCC samples were downloaded from The Cancer Genome Atlas (TCGA) and the Shanghai Ninth People's Hospital. R software packages were used to establish NMF clustering, from which relevant prognostic models were developed. Results: Based on NMF, samples were allocated into 2 subgroups. Predictive models were constructed using differentially expressed genes between the two subgroups. The high-risk group was associated with poor prognostic outcomes. Moreover, multi-factor Cox regression analysis revealed that the predictive model was an independent prognostic predictor. Conclusion: The NMF-based prognostic model has the potential for prognostic assessment of HNSCC.