Mortality prediction in hemodialysis patients using heart rate variability and skin sympathetic nerve activity

Patients undergoing maintenance hemodialysis (HD) face a substantially elevated risk of all-cause mortality, yet robust tools for individualized risk stratification remain limited. This multicenter study developed a predictive model integrating dynamic autonomic nervous system (ANS) markers - heart rate variability (HRV) and skin sympathetic nerve activity (SKNA) - with clinical factors to assess mortality risk. We enrolled 198 HD patients from two Chinese centers between 2021 and 2023, recording HRV/SKNA parameters at baseline, 30 min, and 240 min into dialysis. Over a median follow-up of 34 months, the all-cause mortality rate was 17.7%. Ninety-one baseline features were included in the LASSO-regression model. The final multivariable logistic regression model incorporated six variables (diabetes mellitus, DBP_2h, RMSSD₂₄₀, ΔNnmean₃₀, ΔApEn₃₀ and ΔaSKNA₂₄₀) into the nomogram. The AUC of the nomogram for predicting one-year, two-year, and three-year survival rates was 0.764, 0.749, and 0.805, respectively. The Kaplan-Meier curves for overall survival stratified by nomogram model showed a significant difference between high- and low- risk groups. Internal validation via bootstrap resampling confirmed model robustness, with optimism-corrected AUCs of 0.758, 0.736, and 0.788 for one-, two-, and three-year mortality, respectively. The model demonstrated superior predictive accuracy for cardiovascular mortality (C-index = 0.881) and consistent performance across age and sex subgroups. The proposed model has the potential to predict all-cause mortality in HD patients and may enable earlier intervention and personalized management.