i2b2 Mapped Importer

Java program to take input from various sources and load it into i2b2.

Sources

The aim is to allow any type of source to be used that comes in a per patient format. For example, a CSV file with a row per patient, an XML file with a node per patient, a set of XML files with one per patient, or a database query with a row per patient.

Multiple rows (etc) can be supplied for a single patient.

Outputs

The data can be outputted as either an i2b2 PDO XML file or loaded directlt into an i2b2 database (currently only using Microsoft SQL Server).

Settings

The source and output to used is defined is a YAML file. By default this file is called settings.yaml in the application root directory, but the setings file path can be defined using the -s=xxx or --settingsFilePath=xxx command line arguments.

The settings that can be defined are dataSource and output. Both settings have a field called type that defines the class to be used as a source or output processor. The further fields are specific for the type specified.

Here is an example settings file:

output:
 type: uk.org.briccs.onyxmappedexport2pdo.output.mssql.MsSqlOutput
 server: Server Name
 instance: Instance Name
 database: Database Name
 username: Username
 password: Password
dataSource:
 type: uk.org.briccs.onyxmappedexport2pdo.onyx.OnyxDataSourceSet
 export_directory: data

Available DataSources

There are three data sources at present, with others to be written as the need arises. The available ones are:

• MsSqlDataSourceSet
• MySqlDataSourceSet
• OnyxDataSourceSet

MsSqlDataSourceSet

Settings

• type: uk.org.briccs.onyxmappedexport2pdo.input.sql.!MsSqlDataSourceSet
• Server: The name of the SQL Server
• instance: The name of the SQL Server instance
• database: The name of the SQL Server database
• username: The username required to log on
• password: The password for the above user
• query: The query used to extract the patient data

The query should return a row for each patient, and contain a patient identifier.

Mapping Source

The source attribute list for mappings (see later) should contain one item: the name of the field from the query.

MySqlDataSourceSet

Settings

• type: uk.org.briccs.onyxmappedexport2pdo.input.sql.MySqlDataSourceSet
• Server: The name of the SQL Server
• database: The name of the SQL Server database
• username: The username required to log on
• password: The password for the above user
• query: The query used to extract the patient data

The query should return a row for each patient, and contain a patient identifier.

Mapping Source

The source attribute list for mappings (see later) should contain one item: the name of the field from the query.

OnyxDataSourceSet

Settings

• type: uk.org.briccs.onyxmappedexport2pdo.onyx.OnyxDataSourceSet
• export_directory: The directory where the exported questionnaires are stored

This is a data source written to import questionnaire data exported from the Onyx questionnaire tool.

The tool exports a set of questionnaires into a series of sub-directories, each directory containing the results for all patients for one questionnaire. Each questionnaire folder contains a file called Entities.xml that maps patients to numbered XML files within the same directory (for example 00000001.xml).

Mapping Source

The source attribute list for mappings (see later) should contain two items. The first is the name of the questionnaire folder, the second is the name of the variable element within the patient's numbered XML file.

Proposed Additional Data Sources

Other data sources that might be useful are:

• CSV
• XML using XPATH
• Other relational databases
• JSON

Entity Mapping

Mappings are used to specify which field from the source gets used for which field in the output. These are specified in a YAML file. By default this file is called mapping.yaml in the application root directory, but the mapping file path can be defined using the -m=xxx or --mappingFilePath=xxx command line arguments.

Mappings can be defined for the following entites:

• Patients, including mappings
• Events / Visits, including mappings
• Observations

Currently mappings are not available for other i2b2 entities such as concepts and observers.

Data Fields

The following data fields are available for mapping entities:

• BooleanTrueField
• DataSourceConditionalExistenceField
• DataSourceConditionalExistenceIfNotNullField
• DataSourceConditionalExistenceIfTrueField
• DataSourceConditionalExistenceIfValueOneOfField
• DataSourceEnumerationField
• DataSourceNumericField
• DataSourceTextField
• LiteralTextField
• DataSourceDateField
• YearOnlyDateField

Observation Fields

These fields can be used to create a observations. An observation will only be create if a non-NULL value is returned.

Some observation fields create an observation with a text value, others a numeric value, and others an enumeration value (actually just a text value). Other observations create an observation without a value of any kind. These are called Existence fields.

Date Fields

Most dates in entities are defined as a list of date fields instead of a single date field. If the first date in the list does not yield a date, the next date field in the list is checked and so on. Only if all dates are undefined will the date be set to NULL. In some cases this will cause an error.

Datasource Fields

These fields take their data from the patient row (etc) from the data source. The source attributes of these fields define how the value is located.

Literal Fields

These fields take their values from the mapping definition itself. In these fields the value is either defined by the source attribute or by the field type itself.

The Fields

BooleanTrueField

Always set the value to true.

Mapping

• type: uk.org.briccs.onyxmappedexport2pdo.fields.!BooleanTrueField

DataSourceConditionalExistenceField

A N D T H E R E S T

Patient Entity

Observation Entity

Event / Visit Entity

Extensions