Enhanced Detection of Parkinson's Disease Using XGBoost and Explainable AI: A SHAP-Based Approach

Abstract

Parkinson's disease (PD) is a neurodegenerative disorder that has a major impact on both motor and cognitive abilities as it progresses. Timely and precise identification is essential for successful handling and care. This research introduces a sophisticated method for identifying Parkinson's disease by combining the XGBoost machine learning algorithm with SHapley Additive exPlanations (SHAP), a prominent technique in explainable artificial intelligence (XAI). XGBoost, recognized for its excellent precision and computational speed, is utilized for constructing a strong predictive model with clinical and demographic information. SHAP values are used to explain the model's predictions and give understanding on how each feature impacts the diagnosis. The merging of XGBoost and SHAP not only improves the clarity of the model but also pinpoints crucial biomarkers and risk factors related to Parkinson's disease. This two-pronged method guarantees a great degree of readability and dependability in the diagnostic procedure, potentially resulting in better clinical decision-making and tailored treatment strategies. Our results show that the effectiveness of integrating advanced machine learning methods with XAI tools in the early diagnosis and treatment of Parkinson's disease.