Connectome-based predictive modelling estimates individual cognitive status in Parkinson's disease

Area of Science:

• Neuroscience
• Machine Learning
• Cognitive Science

Background:

• Parkinson's disease (PD) is progressive, necessitating early risk assessment and intervention.
• Cognitive impairment (CI) in PD is often underdetected, impacting daily life and increasing dementia risk.
• Novel tools are needed to detect and interpret CIs in PD.

Purpose of the Study:

• To investigate the potential of connectome-based predictive modeling (CPM) for predicting and understanding cognitive impairment in PD.
• To assess CPM's ability to identify individuals with varying cognitive status in PD.

Main Methods:

• Resting-state functional connectivity data from 58 PD patients were used to train a CPM model.
• The model predicted a global cognitive composite (GCC) score.
• Validation included cross-validation, permutation testing, and stability analyses.

Main Results:

• The CPM model significantly predicted individual GCC scores (r=0.63, p < .05).
• Both positive and negative brain connectivity networks showed significant predictive power (r ≥ 0.58, p < .05).
• The networks differed in their anatomical distribution.

Conclusions:

• A connectome predictive of cognitive scores in PD was identified.
• CPM shows promise for clinical translation in PD for detecting cognitive impairment.
• Longitudinal studies with external validation are required to confirm findings.