A research group at Shaanxi Provincial People’s Hospital employs explainable machine learning on NHANES data to classify obesity into four patterns. They discover compound obesity—high BMI and waist circumference—significantly elevates Parkinson’s disease risk yet paradoxically reduces all-cause mortality in patients, producing validated nomograms for prediction and prognostic assessment.

Key points

• LASSO+RF with SHAP on 51,394 NHANES participants identifies obesity, age, BUN, HDL, AST, smoking, and gender as top PD predictors.
• Compound obesity (BMI ≥24 kg/m² and WC ≥90/110 cm) shows OR≈1.71 for Parkinson’s disease in fully adjusted logistic models.
• Compound obesity paradoxically reduces patient mortality (HR≈0.41) in Cox models; prognostic nomogram achieves AUCROC up to 0.87 for 24-month survival.

Why it matters: This study reveals obesity’s dual role in Parkinson’s risk and survival, offering calibrated AI-driven nomograms for improved early diagnosis and personalized prognosis.

By examining surrogate markers like the TyG index, researchers revealed a U-shaped association with coronary artery disease in type 2 diabetes. Using clinical data and machine learning, the study shows that extreme TyG levels increase risk. Endocrinologists from Tehran University recommend monitoring these values closely to improve cardiovascular outcomes.