In the previous lesson, you constructed a workbook that allowed you to examine the prescription history of a Washington State doctor. In this lesson, you'll examine the second tip you received, about a doctor in Florida. This doctor is also under suspicion for overprescribing lidocaine, which can be used in the illegal drug trade. You'll start by creating a second page and adding the model you created in the last lesson. Then, you'll update it to show the data for the Florida doctor's prescriptions and analyze it to find suspicious patterns.
Create a second page
Each workbook can have multiple pages that allow you to switch between models and analyses. To investigate the Florida doctor, you'll create another page and add the model to it. Then, you'll add the data for the Florida doctor's prescriptions.
- On the ribbon, next to the Washington state tab, click New page.
- Click Excel or CSV.
- Click Browse my computer, browse to the Florida CSV file, and click Open.
In the first lesson, you developed cards and shared a model that had automatically recorded your workflows. In this lesson, you're going to use that same model, and by adding the Florida data, you will create a valid comparison among the cards comparing the lidocaine behaviors of the Seattle and Orlando doctors.
- On the Add To Page pane, click Model.
- Click Content and choose your Lidocaine analysis model.
- Click Add.
The Florida table is added to the data pane but is not yet associated to the cards on the page created by the model you are reusing. The page and card names match the Washington state page used to share as a model. You'll update the model with the data in the next section.
- In the upper right of the ribbon, click Analysis view.
- In the Dataset bubble, click Update.
- In the Dataset table, change the following parameters:
- Choose Dataset: Florida.Table
- X: X_coordinate
- Y: Y_coordinate
- Click Update.
The imported model updates with data from the Florida CSV file you just imported.
- Click Page view.
- On the ribbon, change the name Page 2 to Florida.
- In the Data pane, click Rename dataset and change Florida.Table to Florida.
- Expand the table.
- Click Dataset options and click View Data Table.
The data in the Florida table is organized in the same format as in the Washington state table. The difference is that this table contains 2,584 records, indicating a significantly higher number of prescriptions written by the Florida doctor. The doctor in Washington had written 1,308 prescriptions during the same period.
- Close the table.
You're now ready to explore the Florida data in greater detail as well as in the same manner you inspected the Washington State data.
Identify anomalies in Florida
As in the first lesson, you'll compare and contrast the cards so you can isolate suspicious lidocaine prescriptions.
The doctor is based in Orlando, so a yellow concentration over Central Florida makes sense. Florida, also a warm-weather state with no income tax, is popular among retirees. The low concentrations of heat in the American Midwest can be explained because, like the dual-state residents from Washington State, some retired residents from Florida prefer to retain their home-state doctors, and fill their prescriptions while living in the American Midwest, during the summer months.
- Click the Distribution Heat Map card and zoom in to Central Florida.
Unlike Washington State, the yellow concentrations in Florida are not all centered on cities (in fact, many of these concentrations include some rural areas of Central Florida dominated by farming, citrus, and cattle). In addition, Central Florida has a history with cocaine trafficking.
- On the Treemap by Type card, click each of the Part D drugs and stop on Lidocaine.
The cards change relative to the selected drug. Once the other cards update, the following data patterns are revealed:
- In Distribution Heat Map, lidocaine prescription hot spots generally appear in irregular locations in comparison with other Part D drugs.
- In Cost Bar Chart by Type, the count of lidocaine prescriptions (1,395) is a little over double that of lidocaine/prilocaine (594), which is a more expensive drug that isn't used as a cutting agent with street narcotics. Again, as in lesson one, this doesn't tell us anything we didn't already know; namely, lidocaine is a common and inexpensive drug.
- Distribution by State reflects the states where lidocaine prescriptions were filled. Florida is by far the state with the highest number of prescriptions filled for this doctor, with Oklahoma and Arkansas in second and third ranking. This is not concerning.
- Time Series by Type, however, reveals something you have not yet observed—namely, an unexpected and significant jump in lidocaine prescriptions from this doctor beginning in March 2017.
Pointing your cursor at the lidocaine line activates a pop-up with more specific numbers broken down by weeks instead of months.
- Beginning with 1/1/2017, drag your pointer over the line and stop on the dot after lidocaine makes its dramatic spike beginning on 3/5/2017.
Something happened involving this doctor and lidocaine prescriptions beginning on the week of March 5, 2017. You can see that they jumped from three to 31 prescriptions in one week; you don't know why.
- In Treemap by Type, click lidocaine again to deselect it and show all the Part D drugs on the page.
The card clearly shows that this spike is significantly different from the normal pattern of Part D drug prescriptions from this doctor.
For context and possible confirmation of your finding involving the Florida doctor, you'll switch back to your Washington state page so you can make a valid comparison between the lidocaine histories of the Seattle and Orlando doctors.
- On the ribbon, click the Washington state tab.
- On the Treemap by Type card, click Lidocaine.
Again, as you learned during the first lesson, the Washington State doctor is showing no unexpected spikes in lidocaine prescriptions during the 2015 to 2017 time period as shown on the Time Series by Type card. This contrasts sharply with the Florida doctor, whose lidocaine prescriptions jumped dramatically, and unexpectedly, in March 2017.
Historically, Florida has been an entry point for importing cocaine because of its proximity to South and Central America, which is the major source or origin for this illicit drug. So it makes sense, in a larger context, that the likelihood is greater that a Florida doctor could be involved in prescribing drugs used in narcotics trafficking. Over time, the lethality of cocaine's mixtures has changed, as this article from Sovereign Health discusses, but lidocaine does appear to endure as one of its primary cutting agents.
- Click Save.
Based on information revealed in Time Series by Type, you decide that you do have the justification to flag the Orlando doctor for suspicious lidocaine prescriptions. You feel additional investigation is warranted and you want to share your results with your supervisor.
Share your workbook
Insights for ArcGIS allows you to share your data, page, analysis, and workbook. In your case, you want to share your workbook so you can show and explain your analysis and justification with your supervisor.
Users with Administrator privileges will have access to all workbooks created within their organization, regardless of whether the workbook is shared. When you create and save a workbook in Insights for ArcGIS, an Insights Workbook item is created within your ArcGIS Online content. When you share it from your Content page, it becomes available to your organization. And when sharing, you're providing read-only access; you retain full editing control over your content.
- If necessary, sign in to your ArcGIS Online organization.
- On the ribbon of the home page, click Content.
Your Content page opens with a listing of your items.
- Check the box for Suspicious Lidocaine Prescriptions.
- Click Share.
- In the Share window, check the box for your organization and click OK.
A read-only version of your workbook will now be available for members of your organization with an Insights license.
In this lesson, you relied on the information you created from your first page about Washington State to discover information about a Florida doctor who could be writing suspicious lidocaine prescriptions. You compared the cards so you could make an objective comparison between a doctor in Seattle and one in Florida. You believe you have discovered an indicator which could justify a formal law enforcement investigation involving the Florida doctor.
In completing this lesson, you investigated prescription data for two different doctors in different states and discovered information that prompted you to flag the Florida doctor's activities as suspicious. (In addition, you illustrated and analyzed data about a Washington State doctor who did not need another investigative look.) By using Insights for ArcGIS, you successfully channeled the resources of your office toward a doctor deserving further scrutiny and away from a doctor who didn't. Knowing this information in advance could save your office effort, time, and money.
Insights for ArcGIS could be applied to any data that lends itself to illustration. By geographically enabling data, and creating cards, you can create workbooks that can be shared with coworkers and supervisors regardless of location.
You can find more lessons in the Learn ArcGIS Lesson Gallery.