What's New in I2E?

Introducing I2E 6.0

New Query Editor

I2E 6.0 introduces our new query editor. The interface is the culmination of many developer-years effort to create a new, web-based, easy-to-use replacement for I2E Pro:

Its availability in any modern web browser combined with the fact that your queries are stored on the server means that it’s even quicker and easier to get up and running with the Linguamatics NLP platform. The new query editor has been designed from the ground up to be better looking and easier to get started, and yet still retain all of the features and capabilities previously included in the Java-based editor, which is still available in I2E 6.0.

These changes mean that users wishing to create, improve and share queries are no longer required to have Java installed on their local machines. The redesign also makes it easier to save queries on the I2E server, insert query items, modify properties for multiple items and review the underlying EASL code to see how the query is assembled.

Index Merging

I2E 6.0 provides a new mechanism to convert a collection of small indexes into a single merged index, to simplify index management and improve query performance. The merged index respects any Index Restrictions included in the collection, allowing this feature to be used to create shrink-wrapped subsets or deduplicated versions of existing indexes.

Ontology and Multi-Lingual Updates

The source data was updated for the ChEBI, Diseases, Genes/Proteins, HPO, MedDRA, MeSH, NCI and RxNorm ontologies along with any optimizations required for improved recall and precision.

Several of the Pattern Classes have been updated to improve their utility: for example, numbers containing a dash to denote less-than-zero will now be matched, and “wt%” terms will now be matched in the Measurements class. In addition, there is better normalization for German and French ranges, increased coverage for Spanish range expressions and extra class aliases (for “Pressure”) added for German, French, Spanish, Italian and Dutch.

In addition, the NLP Connector has been upgraded to include better part-of-speech tagging for Dutch, French, Italian and Spanish (and more consistent behavior with German). Chunking rules have also been improved for all languages and third-party NER tools can be plugged into I2E indexing to allow for different terminology matching rules to be used.

Query Updates

New queries have been added to the Resources Query Tree to help with NLP extraction of relationships. In addition, queries in the Healthcare Query Pack have been updated for I2E 6.0 and there is a new version of the Vitals Profile query which extracts more items and runs more quickly. In addition to improvements in extraction of Blood Pressure, Heart Rate, Pulse Rate, Respiratory Rate and Temperature, the query will now also extract BMI, Height and Weight.

Document Processing Improvements

To allow for easier document identification and use with Index Restrictions, I2E 6.0 now uses a specified column in tabular source data as the document identifier. It also includes improved handling of HL7v2 messages.