. Author manuscript; available in PMC: 2018 Mar 1.

Published in final edited form as: SLAS Discov. 2017 Jan 6;22(3):213–237. doi: 10.1177/2472555216682725

Table 1.

Example Approaches to Quantifying Heterogeneity

Approach	Examples	Characteristics
Univariate, Gaussian statistics	mean,²³⁰ standard deviation,²³⁰ z-score,²⁴ skew,²³¹ kurtosis,²³¹ moment²³⁰	Assumes normal distribution, insensitive to subpopulations, no information on type of heterogeneity
Entropy	Quadratic,^{4, 76, 134} Shannon,²³² Simpson,²³² Renyi²³³	Established measures of diversity and information content, only established for univariate data
Non-parametric statistics	KS statistic^{14, 145}	can improve accuracy of results, no assumptions on distribution, no information on distribution shape
Model functions	Gaussian mixture models^{61, 88}	Assumes there is some number of normally distributed subpopulations, can be applied to multivariate data, normal model may not be appropriate
Combined Metrics	PHI^{4, 37}	Model independent, descriptive of heterogeneity
Spatial methods	fractal dimension,²³³ Pointwise Mutual Information (PMI)²¹	No assumption of distribution, leverages spatial interactions, applies to multivariate data
Temporal methods	Temporal distance between robust centers of mass of 2 feature sets,^13,234	Applies to multivariate data, Method developed based on genomic data