About the Oracle Empirica Signal application

Oracle Empirica Signal offers an analysis environment for exploring post-marketing safety data. The product supports pharmacovigilance and risk management activities at pharmaceutical companies and regulatory agencies.

Oracle Empirica Signal generates statistical scores for combinations of products and events in a product safety database, and detects signals (unexpected associations of products and events).

1. Set up the database(s)

The source data may be public data, such as the data in the FDA Adverse Event Reporting System (AERS); Vaccine Adverse Event Reporting System (VAERS); WHO Vigibase ADR (Adverse Drug Reaction) databases; PMDA Japanese Adverse Drug Event Report (JADER) database; or your organization's proprietary product-safety database. Oracle offers data subscriptions for AERS (beginning with extract 2012Q4, FDA refers to the data as FAERS), VAERS and VigiBase data. Argus Mart data can also be used with Oracle Empirica Signal. OCS provides support for incorporating proprietary data sets.

If you have the Manage Configurations user permission and onboarding for the client machine is complete, you can import the database from within Oracle Empirica Signal and configure the Manage Subscription Data Releases feature.

2. Define data configurations

Once an Oracle safety database is set up, you define data configurations in Oracle Empirica Signal to specify the source data available for data mining and how it can be used. For example, if the source data includes both trade names and generic names for products, the data configuration can specify that data mining be performed using generic names only.

3. Perform data mining runs

Once you have created data configurations, you can perform data mining runs. A data mining run extracts data from the source and applies a statistical algorithm to generate signal scores. Oracle Empirica Signal offers different algorithms as options for data mining runs:

• Multi-item Gamma Poisson Shrinker (MGPS)—For MGPS data mining runs, you can indicate a variety of parameters, such as the number of ways in which products and events are combined. For example, a run can generate scores for combinations of up to five values, enabling the study of product interactions and medical syndromes. You can specify a database restriction to specify cases to include in a run.
• Regression-adjusted Gamma Poisson Shrinker (RGPS)—If R and RGPS are installed, you can include computations of RGPS in a two-dimensional MGPS data mining run that does not include subsets.
• Information component (IC)—You can include Information Component as an additional option to an MGPS run.
• Proportional Reporting Ratios (RRR)—You can include computations of PRR and ROR in an MGPS run. PRR computation relates to product-event combinations only, in two-dimensional data mining runs. In two-dimensional data mining runs, the application computes the ROR only for product-event combinations.
• Logistic Regression (LR)—These data mining runs model how the probability of an event appearing in an adverse event report depends on multiple predictors (both demographic factors and the presence/absence of various products), which can be important for obtaining unbiased results in the presence of polytherapy.

4. View the data mining run results

Once a run is completed and published for use by other Oracle Empirica Signal users, you can view the resulting signal scores in tabular or graphical format. You can select criteria to choose results for review and sort the results as needed. You can also choose a variety of options for displaying the table or graph. If a hierarchy is available, representing different levels for product or event terms, you can specify terms at any of these hierarchical levels before reviewing results. For example, you can view the scores for combinations of a particular product and any event in a particular MedDRA System Organ Class (SOC), the highest level in the MedDRA event hierarchy.

If appropriate site options and permissions are set up, you can drill down from run results to view the list of cases involved. For example, the combination of ProductA and EventB has a high signal score and there are 12 cases with that product-event combination. You can drill down to view a list of those 12 cases, and then drill down to detailed information about any of those cases.

You can also print tables or graphs of results and download them for use in applications such as Microsoft Excel (.xls, .csv) or Microsoft Word (.rtf).

5. (Optional) Monitor potential safety signals

Signal review is an optional feature in the Oracle Empirica Signal and Topics application to view disproportionality analysis statistics for successive updates of safety data, allowing monitoring of alerts and evaluating of changes in safety signals over time.

6. Query data and define and run reports

You can query the source data based on criteria that you specify. You can save the query results as a case series, which is a list of cases of interest.

In addition to reports that are delivered with Oracle Empirica Signal, you can construct your own reports by specifying which variables to include as columns and which variables determine report rows. There are two types of reports:

• Regular reports, which are non-interactive and are run against cases in a case series.
• Interactive reports, which are based on a specified query for which you can supply values when running the report.

Note:

A special type of interactive report is the All Cases Summary report, which is used for drug profiles.

7. View drug profiles

From the Drug Profile page, you can view drug profiles, which are presentations of statistical and other information about one or more selected drugs. Depending on how the drug profile feature is set up, you can view existing charts or create your own. Drug profiles are made up of charts, which can be either of the following:

• A graph based on the results of a data mining run.
• A graph or table based on a report output.

A drug profile layout is a saved collection of charts, as well as their placement and display options.

8. (Optional) Track signals with topics

By selecting Topic Management, an optional feature, from the left navigation pane, you can use the Oracle Empirica Topics feature to organize and track tables, graphs, and documents related to particular subject matter. For example, issues identified through various means, such as automated signal detection, literature review, regulatory requests, and so on, can be saved as attachments to a topic.

You can set up topics to follow workflow through different states, such as Initial, Reviewed, and Closed. You can also assign topics to users and make them visible to work teams.