Machine Learning Approach for Risk Prediction of Erosive Esophagitis in a Health Check-up Population in Taiwan

Erosive esophagitis is a kind of GERD (gastroesophageal reflux disease) that potentially develops adenocarcinoma of the esophagus. We collected the data of greater or equal to 20 undergoing upper gastrointestinal endoscopy in a health check-up center between October 2018 and December 2020. We constructed machine learning models using RF (random forest), SVM (support vector machines), MLP Classifier (multi-layer perceptron classifier), and XGBoost and compared them for cross-validation. The risk prediction model identifies high-risk individuals of EE before endoscope examination.