A simulation-based approach for evaluating replicability of Parkinson's disease neuroimaging biomarkers

The translation of neuroimaging research into clinical practice is limited, particularly in the case of Parkinson’s disease (PD), which lacks a reliable MRI biomarker and consensus on which brain imaging features best predict the diesease and its progression. Multiple replication attempts can improve the understanding of replicability and the likelihood of PD biomarkers in neuroimaging studies. I investigate PD biomarker replicability using the Bayesian framework to answer two main questions:

  1. What are the effects of bootstrap on the results of an MRI study? How many times do we find the result in bootstrapped samples?

  2. Can bootstrap inform on the likelihood of replication?

People

Niusha Mirhakimi
Niusha Mirhakimi
MSc student

Department of Neurology and Neurosurgery at McGill University