Development and Validation of Machine Learning Models for Predicting Tumor Progression in OSCC

Area of Science:

• Oncology
• Medical Informatics
• Biostatistics

Background:

• Oral squamous cell carcinoma (OSCC) is a significant global health concern.
• Accurate prediction of OSCC progression is crucial for timely intervention and improved patient outcomes.

Purpose of the Study:

• To develop and validate machine learning (ML) models for predicting tumor progression in oral squamous cell carcinoma (OSCC) patients.
• To identify key clinical, pathological, and hematological features that predict OSCC progression.
• To provide individual risk estimations for OSCC patient outcomes.

Main Methods:

• A retrospective study involving 1163 OSCC patients from March 2009 to October 2021.
• Collection of clinical, pathological, and hematological data.
• Exploration of six ML algorithms, with performance evaluated using accuracy, sensitivity, specificity, F1 score, and AUC. SHAP values were used for feature importance analysis.

Main Results:

• The Logistic Regression model demonstrated superior performance.
• The best performing model achieved a sensitivity of 94.7% and specificity of 55.3% in the development cohort (AUC: 0.76 ± 0.09).
• Neutrophil count was identified as the most predictive feature for OSCC progression.

Conclusions:

• ML models can significantly enhance the clinical prediction of OSCC progression.
• These predictive tools, utilizing basic patient information, can offer individualized risk assessments.
• The findings may assist clinicians in directing targeted interventions for OSCC management.