Predicting Dry Eye Syndrome Patients’ Responsiveness to Magnetotherapy Using Machine Learning
Mengtian Lin
Co-Presenters: Xiangbo Zhang
College: The Dorothy and George Hennings College of Science, Mathematics and Technology
Major: Computer Science
Faculty Research Mentor: Yousef Nejatbakhsh
Abstract:
Dry Eye Syndrome (DES) is a prevalent ocular condition affecting millions worldwide, often leading to discomfort, vision impairment, and reduced quality of life. Magnetotherapy has emerged as a promising non-invasive treatment; however, patient response varies significantly. This study employs machine learning (ML) techniques to predict DES patients' responsiveness to magnetotherapy, enabling personalized treatment approaches.Leveraging large language models (LLMs) for data segmentation and analysis, we extract critical features from patient records, clinical notes, and treatment histories. Using natural language processing (NLP) techniques, we classify patients into responder and non-responder categories based on symptoms, biomarker levels, and demographic variables. Feature selection methods help identify the most influential predictors of treatment success.Multiple supervised ML models, including Random Forest, Support Vector Machines (SVM), and Neural Networks, are trained on structured and unstructured datasets. Performance is evaluated using accuracy, precision-recall, and F1-score metrics. Additionally, model interpretability is enhanced through SHAP (Shapley Additive Explanations) analysis, providing insights into key factors driving treatment responsiveness.The results of this study demonstrate the potential of ML-driven predictive analytics in optimizing DES treatment protocols. By integrating LLM-assisted segmentation with predictive modeling, this research contributes to personalized medicine, improving patient outcomes and reducing trial-and-error approaches in magnetotherapy applications.Keywords: Dry Eye Syndrome, Magnetotherapy, Machine Learning, Large Language Models, Predictive Analytics, Personalized Medicine