 |
|
|
 |
What's New?
VisuaLinks 3.0 is now available!
Since our last issue of the The Linkletter, we have been busy finalizing this much anticipated release. We are very proud of our efforts and confident that our users will be pleased with the breadth and depth of VisuaLinks 3.0.
Our Feature of the Month and Tips and Tricks section highlight just a few of the new capabilities in 3.0. Visit the VAI web site for full details on the 3.0 release.
|
|
|
 |
Here are a few tips and tricks to help you use VisuaLinks quickly and more efficiently.
|
 |
...that you can turn on and off services, menus and toolbar buttons?
From the View menu, choose View, and then choose Configure Tools.
In the Configure Tools window, select the item you want to modify from the tree structure on the left, then check, or uncheck, the options on the right to enable and disable menu items and buttons.
|
|
|
...you can save objects and associations to a database for later processing?
Use the Output menu to save your data.
You can either save to an Access database (MS Windows operating systems only) or you can save to any existing JDBC-accessible database.
In both cases, you first select the attributes to save:
When saving to an Access database, you can save your data to an existing Access database or create a new Access database to store the data.
When saving to a JDBC-accessible database, you simply enter the correct connection information to connect to the database.
VisuaLinks saves each object type and relationship to a separate table with columns that correspond to the attributes that you choose.
|
|
|
...that you can manipulate object groups in the View collectively?
Clicking on the white-space inside a grouped set of objects in the VisuaLinks View selects the group. Icons appear that allow you to manipulate the group as a whole.
The circle icons  are drawing handles that allow you to either move the group as a whole or resize the group when Allow Resize is set. You can enable and disable Allow Resize in the options for the Move mode.
The reset icon  in the upper left corner resets the sizes of objects back to their original size. If you change the size of one or more objects (again, when Allow Resize is enabled), this icon resets the changed objects to their original size. Note that this has no effect on zoomed objects.
Dragging the Rotate icon  rotates the group.
Clicking the details icon  loads the details for all of the data in the group into the Details Panel.
The Plus and Minus icons  zoom the group in and out, respectively. Note that if you zoom in or out on the group, or on a specific object (using that object's zoom icons), Reset will not change the sizes to their original sizes. Zooming and resizing are not the same operation and, though they seem to have a similar visual effect, do not affect objects in the same way.
The merge icon  merges the group in to a single object icon.
|
|
|
|
 |
Parallel Coordinates Placement Setting
Last month we looked at the first of two new placement settings in VisuaLinks 3.0: Centrality. This month we review the second of the new placement settings: Parallel Coordinates.
|
The Parallel Coordinates placement setting is a "strategically" oriented, versus a tactically oriented, view of VisuaLinks data. By this, we mean that this arrangement of data allows you to see large-scale trends in large amounts of data. Tactical perspectives on VisuaLinks data tend to focus on smaller-scale views of data.
 Parallel Coordinates arranges the objects in the View into a kind of grid, with the objects on the left vertical axis and the objects' attributes along the upper horizontal axis. Unique values for each attribute are listed alphabetically under each attribute name. VisuaLinks then draws lines, moving left to right and up and down, as needed, to connect each object with its attribute values. Though this is difficult to explain with words, when seen visually, it very quickly highlights common attribute values.
Whenever you have data loaded into the View, you can use the Parallel Coordinates placement to layout the data. To set up the Parallel Coordinates placement, select the Placements Panel and click the Parallel Coordinates icon.
|
 The Settings and Advanced tabs under the Placements Panel allow you to change the objects and attributes that are displayed as well as modify how the data are displayed in the View.
The Settings tab allows you to add and remove objects and their attributes.
The list boxes on the left show available, though not selected, items. The boxes on the right show selected items.
The yellow and orange arrows pointing left and right allow you to move items between the two list boxes. The yellow arrows pointing up and down moved selected items up and down in the list.
To add an object, select one or more objects listed on the right and click the right-pointing arrow. To remove objects, reverse this process. Add and remove attributes using the same procedure.
|
 The Advanced Tab allows you to control how the data are displayed in the View, as well as to control how the placement responds to mouse clicks.
The line color button allows you to change the color of the displayed line. The Flash Color button allows you to change the color of lines when they are flashed.
The Proportional / Relative check box changes the way the placement uses vertical space, either filling the space completely or not.
The Show Value Labels option turns on a label in the View Area for each distinct attribute value.
The Edit Modes radio buttons change how the View responds to clicking a line with the mouse. In Normal mode, clicking on a line flashes all the lines connecting all of the objects with a related attribute.
In Isolate Mode, all lines not connected to the one clicked are hidden from view, leaving only the line clicked.
In Hide Mode, the line clicked, and all lines connected to it, are hidden from view. This lets you selectively pare back the amount of data displayed in the View.
|
In the image below, we have modified the view to show the ZIP Code, State and Year attributes for SAR objects. As can be seen, there are strong convergences of lines at various values. For instance, we can see that the state of California (CA) is a common value. We can also see that 2002 is a common year.
In the image shown below, we have clicked on the value of 1999 and the lines for the objects that share this value are shown in blue.
The image below shows the effect of enabling the Isolate edit mode. This mode suppresses all value not related to the value clicked. In the image below, we isolated the Year attribute value of 2001. Clicking the Restore button redisplays all values that were previously suppressed.
In the examples shown above, you can perform additional analysis on the objects that best meet your interest.
The Parallel Coordinates Placement allows you to focus on common attribute values and the objects associated with them. It is a powerful way to view your data that allows you to visually spot commonalities and trends.
|
|
 |
Pharmaceutical Prescription Fraud Patterns
During 2004, losses due to Medicare fraud, according to some industry experts, will exceed $20 billion dollars. Much of this can be attributed to physicians billing and coding fictitious filings, inflated claims, and other types of improper actions. With the new prescription drug benefits recently signed into law as part of the Medicare reform act, the program can expect to see even more types and patterns of fraud and abuse.
The following "observations" are derived from a real-world pharmacy data set where much of the information described prescription-related features. Thus, there were fields for amount dispensed, day's supply, various class descriptors, state payments, gender, and other statuses of the claim. Specifically, the names of the patients and doctors were removed for this discussion and the interpretation of the results is subject to further domain expertise.
The model created for this analysis is shown below. The CLAIM represents the overall transaction and supports all the attributes for the record. For each CLAIM, there is a single RECEIPIENT (e.g., the patient) and a single PROVIDER (e.g., the pharmacy) who filled the prescription.
A proactive query into the database (using Summarize) reveals a RECIPIENT with multiple CLAIMS that supports the same prescription. In this case, the prescriptions were all for Hydrocodone/APAP. Hydrocodone/APAP is a medication for relieving pain - which is used in Vicodin. Hydrocodone is habit forming and has been part of many high-profile drug abuse cases in recent times. In the diagram below, there are 4 prescriptions shown for Hydocodone filled by the RECEIPIENT (shown in the red color) at the same pharmacy location. Looking at the dates of each claim indicates a new prescription was filled each week. Drilling down on the details shows the specifics of each CLAIM including the day's supply - indicating that there may be an abusive situation forming in this case.
Another example found in the database shows commonality for CLAIMS. Sodium Chloride is taken as an intravenous solution used to supply water and electrolytes to the body. Additionally, it can be used to mix/dilute other forms of IV medications. Dexamethasone is considered a form of steroid like Cortisone and is typically used to treat inflammation, joint pains, and even lymphoma. In the following diagram, the RECEIPIENT has filled the same combination of prescriptions 3 times over a 5 day period using the same pharmacy. In this case the frequency of the CLAIMS is suspect because the dates are exact for each type of CLAIM filed.
In this next example, we focus on patterns with larger numbers of prescriptions filled for a single recipient. In the following diagram there are many duplicate CLAIMS made for the same type of prescription. Although this is not unusual for many medications that are taken regularly or have a certain number of refills, it is the details of those reoccurring CLAIMS that raise suspicion in this case. In the example below, the details for the cluster for the drug Premarin (a form of estrogen) shows that there are two CLAIMS made on the same day each for a 30 day supply. The cluster showing Vioxx (used for osteoarthritis) also has two CLAIMS made within 1 week of each other for a 30 day supply. Needless to say, many insurance companies won't cover the cost of the additional or duplicate medications without cause.
In this final example, the RECEIPIENT shown has filed a large number of CLAIMS on the same date. The larger group of CLAIMS, shown in the lower part of this diagram, were all filed on the same day. In fact, there are 12 different prescriptions filled on 05/01 representing over $1000 in co-pay costs. This type of "spike" in CLAIMS processing should be reviewed to determine if there are any abuses occurring in the system. A sample of the types of prescriptions filled is shown below:
Ambien is used to treat insomnia
Gemfibrozil is used for lowering cholesterol and triglycerides
Isosorbide helps prevent chest pain (angina)
Norvasc is also prescribed for angina and high blood pressure
Prevacid treats heartburn
Sustiva is an inhibitor used for HIV treatment
Videx EC is a form of HIV medication
Zestril is used for treating high blood pressure
Zoloft is an antidepressant
There were several other types of patterns exposed in this data set, including the utilization of multiple pharmacies to fill orders within a fairly short time frame - which might be indicative of circumventing protections (checks and balances) within a system. Additionally different combinations of prescriptions, frequencies, dosages, dates, locations, and other variables has lead to the discovery of other patterns not shown here.
|
|
|
 |
|
