Linear predictive coding electroencephalography algorithms predict Parkinson's disease mortality using out-of-sample...

Area of Science:

• Neuroscience
• Biomedical Engineering

Background:

• Parkinson's disease (PD) mortality prediction is difficult due to patient heterogeneity and lack of reliable prognostic markers.
• Accurate mortality prediction is crucial for patient management and clinical trial design in PD.

Purpose of the Study:

• To evaluate the efficacy of the Linear Predictive Coding EEG Algorithm for PD (LEAPD) in classifying 3-year mortality in PD patients.
• To assess the correlation between LEAPD indices and time to death in PD patients.

Main Methods:

• Resting-state electroencephalography (EEG) data from 94 PD patients were analyzed.
• The LEAPD algorithm was used for binary classification of 3-year mortality and correlation analysis with survival time.
• Leave-one-out cross-validation and out-of-sample testing were employed to validate the model's performance.

Main Results:

• Several EEG channels achieved 100% accuracy in mortality classification with robust performance observed from five channels.
• LEAPD indices showed significant correlations with time to death (Spearman's ρ ranging from -0.59 to -0.86), independent of clinical factors.
• Out-of-sample testing demonstrated a mean accuracy of 83% for mortality classification.

Conclusions:

• LEAPD offers a robust method for classifying mortality risk in Parkinson's disease using resting-state EEG.
• LEAPD indices serve as potential continuous neurophysiological biomarkers correlating with survival duration in PD.