Predicting Parkinson’s disease severity using support vector regression based on Bayesian optimization

Based on multimodal data from public datasets， including demographic characteristics， clinical features and imaging features， a support vector regression model optimized through Bayesian optimization was proposed to accurately predict the severity of Parkinson’s disease. Experimental results demonstrated that the model not only exhibited high accuracy in predicting Parkinson’s disease severity but also showed significant explanatory power. Through feature importance analysis， the key features that contribute most significantly to the predictive model were effectively identified， which not only provides a solid scientific basis for clinical management and treatment decisions in Parkinson’s disease， but also opens up new research perspectives for in-depth exploration of the pathological mechanisms of this disease.