Function introduction

High-performance NLP components

CLAMP components are built on proven methods in many clinical NLP challenges including the I2B2 clinical NER (2009/2010-#2), SHARE/CLEF (2013-#1), SemEval2014 UMLS encoding (#1).

Machine learning and hybrid approaches

Depending on the task, users can train their own model for the machine learning based components of CLAMP and evaluate custom models using a custom corpus.

Annotation and corpora management

Users can import clinical text corpora into the CLAMP workspace and annotate files using the built-in annotation tool that can be utilized in CLAMP projects, both as training and test datasets.

Customized pipelines

CLAMP allows building the NLP pipelines by offering all the requisite components such as named entity recognition, assertion, UMLS encoder, and component customizations.

Knowledge sources and sample clinical text

All the knowledge resources required for CLAMP components like dictionaries, section header list, or medical abbreviation list are provided along with it.

Interoperability and Scalability

CLAMP is built on the UIMA framework, and is therefore compatible with other systems such as cTAKES. Further, CLAMP also utilizes the cTAKES’ type system for lower linguistic level annotations.