similarityscore

This function computes person-based similarity index score, based on inter-individual variability.

Person-Based Similarity Index (PBSI) Function: [PBSI,SIMS]= similarityscore(M)
This function computes both PBSI and intrapair similarity scores for each individual, for a specific dataset (either structural or functional). It is based on inter-individual variability computed across multiple brain variables.
The conceptual unit of the PBSI is the individual person’s brain structural (or functional) profile which considers all relevant measures as features of a single vector. It represents a metric of the similarity between the brain profile of an individual study participant and that of all other participants.
This function is also very useful prior to clustering participants, as it reveals the degree of similarity among individuals for a specific phenotype (using the SIMS output).
Input:
M = matrix p*n, where p (rows, p must be >=2) are the participants (i.e., members of the same group) and n (columns) are the variables for a specific phenotype (example: all cortical thickness measures)
Outputs:
PBSI = vector p*1: 1 value per participant reflecting: the person-based brain similarity, based on p-1 participants.
SIMS = symmetric matrix p*p, reflecting the correlation between participants' profile (based on the n variables)
a = outlier participants (PBSI with a deviation from more than 4 STD)

This function ignores missing values.

Reference for this function is:
Gaelle E. Doucet, Dominik A. Moser, Amanda Rodrigue, Danielle S. Bassett, David C. Glahn, Sophia Frangou (2018) Person-based brain morphometric similarity is heritable and correlates with biological features. Cerebral Cortex.