Release 64

Features

New Login Page

As part of our work towards supporting Identify Provider (IdP) /Single Sign-On (SSO) authentication, we converted our login page to the new version of our front-end technology. The login page now has a more modernized experience of the user entering their e-mail address and then password as a secondary step.

User Mobility Calculation Change

The User Mobility report attempts to find users who have unusual patterns around where they work. A hospital will be made up of several different departments, and nurses will generally have a home base where they perform most of their work. The main focus of the analysis is to highlight providers who switch departments during their shifts. 

The original version of the report counted the number of times a provider worked in each department. Drawbacks of this method:

• Providers who worked in a single department, but worked much more than their peers, would be flagged as unusual.
•  Providers who worked in a single department, but had a single event in another department were flagged as being exceptionally unusual.
•  Float nurses, who work shifts in a different departments, would be flagged as unusual, despite their activity being completely normal.
•  Entire departments, where only a handful of providers would work, were being given very high scores.

The new version of the mobility report attempts to highlight providers who switch department during their shifts. This analysis has three main components:

• Assigning provider shifts: Float nurses will typically spend an entire shift in a single department. For that reason, we do not want to flag them as unusual if they start a new day in a different department than the one they ended the previous day in. 
•  We are therefore only looking for department switches within a shift. To split the activity of a provider by the shift in which they were working, we group together all activity for a provider which falls within a 8-hour cool-off period.
•  Modeling the department transition probabilities: The previous bullet allows us to now calculate the probability of a provider transitioning from one department to another and we can also include some history in these calculations (where was that provider before making the department switch).
•  For example, if a provider has just had 2 events in the NICU, what would be the probability of their next event occurring in the PACU? And then, what is the probability of the provider going straight back to the PACU, given that they had two preceding events in the PACU and one in the NICU?
•  We calculate the probability of switching from one department to another by considering which departments the last 5 events were in for that provider.
•  These models are created using a year's worth of data. This allows us to get smarter over time, as we get more data from our newer customers, while not including very old data in the calculations for our existing customers.
•  Now that we have the transition probabilities saved for each hospital, we can pull down these models and analyze the data to find outliers.

Did we address the issues with the original version of the report?

• We are only looking for transitions between departments, so if a provider has only worked in a single department, they would be given a score of 0.
•  The more unusual switches you have, the higher your score, we will therefore be highlighting patterns of behavior rather than one-off events.
•  So long as a float nurse stays in a single department for an entire shift, they will not be flagged as unusual.
•  In small departments, if the providers do not perform events in unusual departments, these providers will not be flagged as unusual.

No visualization changes were made to the report, but we did adjust the natural language in IRIS to convey how the analysis is being performed.

Med Trends Natural Language

There was an issue when a provider had unusual behavior for multiple form factors of the same drug. We enhanced the natural language description in IRIS to also list the form factor. The deep linking will also automatically include the report filtered by the correct form factor. 

New Column Added to Summary Report Downloads

Previously, "Case Start Time" in the Run Report Summary Report download was the time of the first event. This as compared to the Reconciled Time is not an accurate depiction of how long it took to close, as the data for the event could have been uploaded days, weeks, or months after the event actually occurred. We changed the following:

•  Changed "Case Start Time" to be equal to the "created date/time" of the event summary record itself. This will be the first time that the customer could have seen the event summary in the application.
•  Created a new column "First Event Time" to have the date/time of the first event, so that that information is still available.

Upgrades & Fixes

• Continued work to further speed up the Audit page.
• There were a few issues where the PDF that is download using the 'Download PDF' feature on the various analytics were being cutoff. This has been resolved.