Explainable computational framework for subtype classification and biomarker identification in head and neck squamous cell carcinoma using machine learning
- 1Department of Sciences, Amrita School of Engineering, Amrita Vishwa Vidyapeetham, Chennai, India
- 2School of Computer Science and Engineering, Vellore Institute of Technology, Vellore, Tamil Nadu, India
- 3Department of Computer Science and Engineering, Amrita School of Computing, Amrita Vishwa Vidyapeetham, Chennai, India
Abstract
The precise classification of basal, classical, atypical, and mesenchymal subtypes of Head and Neck Squamous Cell Carcinoma (HNSCC) is crucial for advancing precision oncology. Our research aims to develop an explainable subtype classification model for HNSCC.These molecular subtypes have been previously defined; reliable and meaningful computational methods for the classification of molecular subtypes and prioritization of molecular biomarkers for each subtype do not exist. We developed an explainable artificial intelligence (XAI) model that classifies molecular subtypes based on combined transcriptomic profiles from The Cancer Genome Atlas (TCGA-HNSC) and Gene Expression Omnibus (GSE65858). SHapley Additive exPlanations (SHAP) were used to quantify the importance of genes for model predictions. Subsequently, genes with high predictive importance were analyzed in protein-protein interactions using STRING, and hub genes were identified by Maximum Clique Centrality (MCC) and used for the identification of candidate biomarkers for each of the subtypes. We employed machine learning methods such as Random Forest (RF), Logistic Regression (LR), Decision Tree (DT), AdaBoost (AB), Support Vector Machine (SVM), K-Nearest Neighbours (KNN), Gaussian Naive Bayes (GNB), XGBoost (XGB), and Gradient Boosting (GB). The XGBoost model achieved an accuracy of 95% after hyperparameter tuning and cross-validation. Biologically relevant subtype-specific biomarkers associated with each molecular subtype were identified. To highlight the potential biological functions of these biomarkers, pathway analysis was performed. Further assessment of these genes, such as survival analysis and druggability, showed promise for the development of targeted therapies.
Related
- Class-Conditioned Diffusion Augmentation and Explainable In-Context Ensemble Learning for Imbalanced Maintenance Action Classification in Aircraft Systems
- On the Computational Complexity of Hidden Markov Model Identification
- An integrated explainable AI framework with counterfactual optimization for decision support
- GCUL: Ambiguity Identification in Text Emotion Classification via Cluster-Guided Learning
- A Hybrid Structural Modelling and Explainable Machine Learning Framework for Human–AI Strategic Decision Support Systems
BibTeX
@article{rajan2026explainable,
title = {{Explainable computational framework for subtype classification and biomarker identification in head and neck squamous cell carcinoma using machine learning}},
author = {Anagha Rajan and Christopher Leela Biji and Mangalam Goutam Sinha and Iyappan Ramalakshmi Oviya},
journal = {PLOS One},
year = {2026},
month = nov,
doi = {10.1371/journal.pone.0360612},
url = {https://doi.org/10.1371/journal.pone.0360612},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - Explainable computational framework for subtype classification and biomarker identification in head and neck squamous cell carcinoma using machine learning
AU - Anagha Rajan
AU - Christopher Leela Biji
AU - Mangalam Goutam Sinha
AU - Iyappan Ramalakshmi Oviya
PY - 2026
DA - 2026/11/04
JO - PLOS One
DO - 10.1371/journal.pone.0360612
UR - https://doi.org/10.1371/journal.pone.0360612
AB - The precise classification of basal, classical, atypical, and mesenchymal subtypes of Head and Neck Squamous Cell Carcinoma (HNSCC) is crucial for advancing precision oncology. Our research aims to develop an explainable subtype classification model for HNSCC.These molecular subtypes have been previously defined; reliable and meaningful computational methods for the classification of molecular subtypes and prioritization of molecular biomarkers for each subtype do not exist. We developed an explainable artificial intelligence (XAI) model that classifies molecular subtypes based on combined transcriptomic profiles from The Cancer Genome Atlas (TCGA-HNSC) and Gene Expression Omnibus (GSE65858). SHapley Additive exPlanations (SHAP) were used to quantify the importance of genes for model predictions. Subsequently, genes with high predictive importance were analyzed in protein-protein interactions using STRING, and hub genes were identified by Maximum Clique Centrality (MCC) and used for the identification of candidate biomarkers for each of the subtypes. We employed machine learning methods such as Random Forest (RF), Logistic Regression (LR), Decision Tree (DT), AdaBoost (AB), Support Vector Machine (SVM), K-Nearest Neighbours (KNN), Gaussian Naive Bayes (GNB), XGBoost (XGB), and Gradient Boosting (GB). The XGBoost model achieved an accuracy of 95% after hyperparameter tuning and cross-validation. Biologically relevant subtype-specific biomarkers associated with each molecular subtype were identified. To highlight the potential biological functions of these biomarkers, pathway analysis was performed. Further assessment of these genes, such as survival analysis and druggability, showed promise for the development of targeted therapies.
SN - 1932-6203
ER -