Research on acupuncture treatment effect prediction for migraine based on stacking

As a method widely used in the treatment of migraine in clinic, the acupuncture is effective, but there are still great differences between the treatment effects of different patients. To improve the ability to predict patient treatment outcomes for migraine patients, Stacking ensemble learning method is used to construct a predictive model for the efficacy of acupuncture treatment for migraines. Collect a large amount of migraine patient data, including Visual Analogue Scale (VAS), Migraine Specific Quality of Life Questionnaire (MSQ), and patient symptoms, and perform data preprocessing and feature engineering. The Support Vector Machines (SVM), Random Forests (RF), Gradient Boosting (GBDT), CatBoost, Extreme Gradient Boosting (XGBoost) and other machine learning methods is used to establish the prediction model of the effect of acupuncture on migraine. These models are arranged and combined as base learners, and the Multi-layer Perceptrons (MLP) are used as meta learners to construct Stacking models. The hyperparameters are optimized through grid search methods to further improve prediction performance. The experimental results show that the final Stacking model with the best performance achieved a prediction accuracy of 93.16%, while the accuracy, recall and the F1 scores are also above 93%. The prediction results of these models were validated through Magnetic Resonance Imaging (MRI) data, further confirming their reliability and effectiveness. The method provides an important reference and support for clinical decision-making and acupuncture treatment of migraine.