About Oracle Thesaurus Management System (TMS)

Pharmaceutical companies that develop new drugs are required to produce reports that document the effects of the drug. However, the data used to produce these reports, including information on health conditions and medications, can be hard to analyze because the terminology is inconsistent. Even when a company uses a dictionary of appropriate terms provided by an organization such as the World Health Organization, data collected during the study may be inconsistent with its terminology for a variety of reasons:

• Medical personnel may use terminology different from that of the dictionary.
• Companies may have developed new drugs since the dictionary's creation.
• A study may require more precise terminology for a health condition than is available in the dictionary.
• Medical personnel may misspell drug names or health conditions.
• Data entry personnel may misinterpret or mistype the data.

TMS can transform this data into a consistent terminology that statisticians can analyze more easily.

TMS refers to an item of data initially entered or loaded into the system—drug or health condition information or other data item—as a verbatim term (VT). TMS maps verbatim terms to dictionary terms in a process called classification. TMS automatically classifies verbatim terms that directly match a dictionary term, and enables you to classify variant or misspelled terms manually. You can choose to have future occurrences of the same verbatim term automatically classified the same way, or you can choose to require manual classification, with the previous classification presented. You can also add terms to vendor-supplied dictionaries and modify your own terms at any time.

This section includes:

• Usage Models
  
• Dictionary Definition and Customization
  
• The API
  

Usage Models

TMS serves as the single terminology repository for clinical studies, and can be made available to all applications used in studies. Particular applications, or external systems, are more or less compatible with TMS. Depending on the degree of compatibility, TMS can be installed in different degrees of integration with the external system(s).

The more fully TMS integrates with your clinical study application, the better it serves you. However, you can still benefit from TMS if your other software permits only partial integration, or even no integration.

This section includes:

• Full Integration
  
• Partial Integration
  
• No Integration
  

Full Integration

Full integration requires the installation of TMS objects in the external system, a level of integration that is recommended if the external system is running in a stable global Oracle environment. In full integration, TMS maintains external data in both the tms_source_terms table and the tms_vt_omissions table.

A fully integrated system benefits from the full range of TMS functionality. It feeds source terms to TMS with contextual data you specify (such as Patient and Document Number) so that if you reclassify or declassify a term in TMS, TMS can send information about each affected source term back to the source data system.

In TMS, you run Autoclassification, manually classify remaining terms (omissions), assign Actions, and reclassify or declassify as necessary. You specify the information you want to derive from TMS for each source term, and TMS sends that data to the source data system associated with each source term.

Partial Integration

If full integration is not desirable—for example, if space is an issue—you may want to set up TMS with only partial integration to your external source data system. In partial integration, TMS maintains external data in the tms_vt_omissions table only. It does not store any data in the tms_source_terms table.

The external system feeds data to TMS, and TMS returns derived data and omissions. You can classify terms manually and perform all other functions within TMS. However, because the tms_source_terms table is not used, TMS has no external system information (such as patient ID or document number) associated with source terms and therefore cannot associate derived terms with source terms after the initial classification.

See Partial Integration for more information.

No Integration

Even if TMS has no knowledge of your external system, you can use TMS as a dictionary repository. You can use TMS to access, add, and modify dictionary terms. You can take advantage of TMS's flexible hierarchical dictionary structure and data validation. However, TMS functionality in this scenario is limited in the following ways:

• TMS cannot receive terms directly from the external system, but you can load them or enter them manually in TMS.

• TMS cannot handle omissions.

• The external system must handle the impact of reclassification and de-classification on derived data.

• High-level classification is not available.

Following is one scenario for using TMS with no integration to the external system:

1. Check for matches between the external system's source data and the TMS repository.
2. Load or create VTAs in TMS for verbatim terms in the external system that did not map to a term in the TMS repository.

The no-integration approach is recommended if the external system does not run on an Oracle database.

Dictionary Definition and Customization

TMS supports the inclusion of any number of dictionaries in its repository. You can use dictionaries supplied by a vendor such as WHO (an external dictionary) and/or convert a legacy company dictionary. You can customize external dictionaries by adding company terms. You can customize dictionaries in multiple ways, all available simultaneously, by using TMS domains.

This section includes:

• Flexible Definition
  
• Customization
  

Flexible Definition

You can create a dictionary in which terms inherit the relations of the levels to which they belong, or one in which you store all the terms on a single level and use the relation itself to define the connection between the terms. For dictionaries with multiple levels, you can define any number of levels in a given dictionary; you can group levels, and you can create a hierarchy of any shape or degree of complexity. You define both ends of the relation between each pair of related levels, specifying:

• whether the relation is optional in each direction, and in some cases, whether the relation is mandatory in each direction

• whether a single term in one level can map to more than one term in the other level, in each direction

• whether the parent-level end of the relation can have multiple links, whether a term in the child level must have a Primary Link to a single term in the parent level

• the currency of viewable data. Base dictionaries contain all dictionary information that is currently not expired in the TMS repository, while virtual dictionaries can represent the state of a dictionary at a particular point in time.

See Defining and Loading Dictionaries for more information.

TMS automatically performs validation to ensure that links between terms do not violate the level relations you have defined.

Customization

You can add company terms to external dictionaries if, for example, you need to use the name of a new drug that is not included in the external dictionary.

If you need several more specific terms instead of one more general term provided by the dictionary, you can mark the external dictionary term as Unapproved to ensure that it is not used for classification, forcing the use of the more specific company terms instead.

Different studies or projects may have different terminology needs. TMS addresses this issue with the concept of user-defined TMS domains. You associate external dictionaries with TMS domains and customize them differently in each domain, creating TMS domain-specific terms and relations. See Defining and Loading Dictionaries for more information on defining and using domains in TMS.

The API

TMS includes PL/SQL packages to accomplish all TMS tasks, providing the ability to interface with many external applications. Your external system calls these packages instead of manipulating the database directly, thus avoiding database corruption. The API is documented in the Oracle Thesaurus Management System Technical Reference Manual. Contact Oracle Support to receive a copy.

The following table shows a subset of the API packages, and the purpose of each one.

Table 1-1 Sample API Packages and Their Purpose

Package	Purpose
tms_user_classification	Creates accepted or misspelled verbatim term assignments (VTAs), which map the corresponding correct or misspelled verbatim terms to dictionary terms.
tms_user_action	Enables you to specify comments for unclassified VTs, which can be propagated to the external system's discrepancy management system.
tms_user_reclassification	Reclassifies VT to another dictionary term or declassifies the VT.
tms_user_fullautocode	In fully integrated external systems, this package matches a new or updated verbatim term with a dictionary term, and returns values.
tms_user_autocode	In partially integrated external systems, this package matches a new or updated verbatim term with a dictionary term.