Labs - Action Sequences Metric

Action Sequences Dashboard

The Action Sequences metric allows you to investigate the order in which actions are performed on a patient. More specifically, you can look for patterns of interest in the order of actions recorded for a given patient. This includes tracking the order of a medication's dispense (D), administration (A), waste (W), return (R), and transfer (T). 

As an example, this article will explore the sequence of actions related to the administration of Morphine at a facility. To map related actions, the letter S will be used to denote the start of the patient-medication-user sequence and the letter E will denote the end of a sequence. A common order of events for Morphine might be a single dispense followed by an administration, represented by S > D > A > W in our mapping. The screenshot below illustrates the most common action sequences for Morphine at our test facility.

The red dot labeled S denotes the start, the dark blue dots represent the actions taken, and the light blue dots, labeled E, represent the end of the sequence. The thickness and color of the lines connecting the actions show how common the sequence of actions is in your data. To see the probability of a given path, hover your mouse on any of the lines. For ease of use, we have truncated the no. of actions shown in a sequence to a maximum of 5. 

Finding Unusual Sequences 

Based on your calculated sequence mapping and frequency assessment, this metric will look for the most unusual sequences involving the desired drug (e.g., Morphine).  To determine which sequence is unusual, emphasis will be placed on the length of the sequence. The screenshot below illustrates an example of the top ten most unusual sequences from a facility based on length and how often they appeared (Count).

Another method for identifying unusual sequences emphasizes uncommon transitions between actions. For example, it may be unusual for a dispense to be recorded after a return or for a dispense to be the last action taken in a string of transactions. The screenshot below displays an example of the ten most usual sequences from a facility based on transaction order and how often they appeared (Count).

Finding Unusual Users

Utilizing this data, you can assess unusual sequences to identify users who create unusual patterns with more frequency than their peers. Then each provider is assigned and ranked with an outlier score. Review the screenshot below to see an example of a provider, Christopher, and review his ten most unusual event sequences.