Developing Robust Predictive Models for Head and Neck Cancer across Microarray and RNA-seq Data.

Area of Science:

• Genomics and Bioinformatics
• Oncology
• Computational Biology

Background:

• Understanding transcriptomic patterns in head and neck squamous cell carcinoma (HNSCC) is crucial for early diagnosis and improved treatment.
• Integrating multi-study data is essential to identify consistent gene expression signatures for HNSCC detection, especially at early pathological stages.

Purpose of the Study:

• To develop robust predictive models for HNSCC disease status using feature integration and heterogeneous ensemble modeling.
• To apply these models for discriminating early-stage HNSCC from normal samples using transcriptomic data.
• To create a user-friendly software tool for HNSCC researchers to analyze gene expression datasets.

Main Methods:

• Utilized feature integration and heterogeneous ensemble modeling techniques.
• Applied models to both microarray and RNA-sequencing (RNAseq) datasets.
• Developed a software tool with a graphical user interface for predictive modeling.

Main Results:

• Several models achieved high performance, with Matthews Correlation Coefficient (MCC) and Area Under the Curve (AUC) values exceeding 0.8.
• Models showed encouraging results in discriminating early pathological stage HNSCC from normal RNAseq samples.
• The integrated software tool allows researchers to leverage transcriptomic features and ensemble models for new HNSCC datasets.

Conclusions:

• Robust transcriptomic-based models can effectively predict HNSCC status and aid in early detection.
• The developed computational tool provides a valuable resource for HNSCC research, facilitating the analysis of gene expression data.
• This approach enhances the ability to identify HNSCC through consistent gene expression signatures.