In order to facilitate a convenient utility of CLAMP-Cancer, a series of pipelines that could be
directly adopted in common clinical applications are pre_built and displayed in
PipelineLibrary. Users can directly drag one of them (e.g., smoking_status,
) from the PipelineLibrary and drop it under My Pipeline (see figures below). The required NLP
components of these pipelines are already configured. CLAMP-Cancer
allows you to customize each of these components to fit your needs.
First, you need to
import your files; for more information go to
"Import input files”
section.
Depending on what your use case is, the current built-in pipelines are divided into the
following categories:
General: automatically annotates concepts and their attribute for general
use, including:
CLAMP-Cancer-ner: annotates the disease, procedure and medication concepts
CLAMP-Cancer-ner-attribute:
annotates the attributes of disease (e.g., body
location of a
disease), lab procedure (e.g., value of a lab test ) and medication (e.g., dosage of a
medication) concepts
Disease-attribute:
annotate the attributes of diseases, including body
locations (e.g.,
left atrium), severity degrees (e.g., mild, severe) and uncertainty (e.g., probably). Lab-attribute:annotates the
attributes of lab procedures Medication-attribute:
annotates the attributes of medications
Disease_symptom:
automatically annotates symptoms of diseases, including:
Bleeding_extraction:
annotates bleeding symptoms
Colorectal_cancer:
annotates symptoms of colorectal cancer
Behavior:automatically annotates behaviors of patients , including: Smoking_status:
annotates whether or not the patient is in a smoking status, and
whether the patient has a smoking history.
The figure below illustrates an example of using the disease-attribute pipeline in our
pipeline
library to annotate attributes and their relations with diseases.