Response to venom immunotherapy: Exploratory retrospective machine learning clustering analysis

Background: Hymenoptera venom immunotherapy is an established treatment for severe allergic reactions, aiming to modulate the immune response and reduce allergen sensitivity. However, traditional methods such as skin tests and specific IgE quantification often lack precision and fail to capture the multidimensional nature of clinical data. Objective: We investigated the relationships between wheal surface area, specific IgE levels, and patient age in allergic reactions to Hymenoptera venom, assessing immunotherapy effects before and after treatment. Methods: Data were retrospectively collected from 30 patients who underwent intradermal testing before and after immunotherapy. Wheal surface areas were measured using the semiautomated method (SAM), and specific IgE levels via immunoassays. K-means clustering, an unsupervised machine learning technique, was applied to identify patient subgroups on the basis of an integrated analysis of the 3 variables. Data normalization ensured comparability across different units. Results: A positive correlation between wheal surface area and specific IgE was observed before and after treatment, both showing reductions after immunotherapy. Age showed no significant influence. Clustering revealed two consistent profiles: response and partial response. The Müller scale confirmed clinical improvement with reduced reaction severity. Conclusion: Immunotherapy reduces allergic response, as shown by decreased wheal size and IgE level. The integration of SAM and machine learning enables robust analysis of clinical data, supporting personalized allergy management.