Predicting Chronic Hyperplastic Candidiasis Retro-Angular Mucosa Using Machine Learning.

Area of Science:

• Oral pathology
• Mycology
• Computational biology

Background:

• Chronic hyperplastic candidiasis (CHC) is a rare oral infection presenting as white/red mucosal patches.
• Lesions can be misdiagnosed as leukoplakia or erythroleukoplakia due to similar appearance.
• Predominant CHC locations include the tongue, retro-angular, and buccal mucosa.

Purpose of the Study:

• To investigate the influence of anatomical location, specifically retro-angular mucosa, on CHC development.
• To develop a machine learning (ML) model for predicting CHC occurrence based on risk factors.
• To enhance diagnostic accuracy and inform treatment strategies for CHC.

Main Methods:

• Utilized Gradient Boosting Regression (GBR), a machine learning algorithm.
• Classified CHC lesion locations by analyzing key risk factors.
• Evaluated model performance using MSE, RMSE, R-squared, and MAE.

Main Results:

• The ML model successfully predicted CHC occurrence in the retro-angular mucosa.
• The GBR approach demonstrated high accuracy in classifying lesion locations.
• Performance metrics confirmed the robustness and reliability of the classification method.

Conclusions:

• Machine learning offers a valuable tool for predicting CHC location, especially in the retro-angular area.
• The proposed ML technique aids in improving diagnostic accuracy for CHC.
• This research provides insights for both clinical diagnosis and further research into CHC.