Machine Learning-Based Prediction of Free IgE Concentration in Allergic Rhinitis Patients Treated with Allergen Immunotherapy and Omalizumab

Abstract

Free immunoglobulin E (IgE) concentration is a key biomarker for allergic diseases. Prediction of free IgE concentration can help clinicians diagnose and monitor allergic diseases more effectively. In this study, we used machine learning to predict free IgE concentration in the blood serum of patients with allergic rhinitis who received allergen immunotherapy co-administered with omalizumab. The predictors for free IgE concentration were the number of visits for treatment and baseline checking, and treatment groups (1) omalizumab/ragweed, (2) omalizumab/placebo, (3) placebo/ragweed, and (4) placebo/placebo. Several machine learning algorithms (MLA) were trained with the immunotherapy dataset imported from Immune Tolerance Network (ITN) TrialShare into the Orange data mining platform. The decision tree algorithm model amidst the list of MLAs trained and tested was the best performing model for predicting free IgE concentration, with an R-squared of about 0.6. This study demonstrates that machine learning can be used to predict free IgE concentration with high accuracy. This prediction model could be used to help clinicians diagnose and monitor allergic diseases more effectively.