Multimodal artificial intelligence-based pathogenomics improves survival prediction in oral squamous cell carcinoma

Area of Science:

• Oncology
• Bioinformatics
• Artificial Intelligence

Background:

• Oral squamous cell carcinoma (OSCC) poses a significant health challenge with variable patient prognoses.
• Accurate prognostic prediction is crucial for tailoring treatment strategies and improving patient outcomes in OSCC.

Purpose of the Study:

• To develop a novel prognostic algorithm for OSCC by integrating pathogenomics and artificial intelligence (AI) techniques.
• To identify key predictive features for OSCC survival outcomes using machine learning and deep learning.

Main Methods:

• Utilized comprehensive clinical, genomic, and pathology data from 406 OSCC patients in The Cancer Genome Atlas (TCGA) dataset.
• Performed gene expression, principal component, gene enrichment, and feature importance analyses.
• Applied five machine learning/deep learning algorithms: Random Survival Forest, Gradient Boosting Survival Analysis, Cox PH, Fast Survival SVM, and DeepSurv for survival prediction.

Main Results:

• The multimodal AI model demonstrated superior performance compared to unimodal models across all tested algorithms.
• Multimodal models achieved higher c-index values, indicating improved predictive accuracy for patient survival.
• Feature selection using important features further enhanced the predictive power of the multimodal models.

Conclusions:

• Pathogenomics combined with AI-based techniques offers a powerful approach for enhancing prognostic prediction accuracy in OSCC.
• The developed prognostic algorithm holds potential for guiding personalized treatment strategies in OSCC patients.
• This study highlights the synergy between multi-omics data and advanced AI for advancing cancer prognostics.