Table 3.

The actionable measurement gap separation method for finding and removing a confounding bias in laboratory test EHR data.

Measurement Gap Separation Method
Step	Action	Motivation
1	Plot a histogram of the frequency and measurement gap in log-log coordinates.	The histogram provides a method to visually examine the laboratory tests measurement dynamics.
2	Examine the modality of the plot; looking for multi-modality.	If the histogram is multi-modal, it may imply a difference in patient health states or a healthcare process bias.
3	If there are multiple peaks, define a measurement gap threshold to separate the peaks.	This separation defines multiple settings for the EHR experiment, creating sets of homogenous data points with respect to their measurement gaps.
4	Perform the EHR experiment separately for each setting.	Separately performing experiments for different settings may remove confounding bias.