NLP Clinical Ontology 101

Clinical ontologies are the most underrated tools in healthcare NLP. In this lightning talk targeted for healthcare NLP Beginners, Marcelo will give an introduction about which are the most important healthcare ontologies, why they are important, how they connect to each other, and some examples on how to use them in healthcare Natural Language Processing.

The talk will cover, amongst other topics:

1) The ICD diagnosis coding system, how it is organized, and how to interpret and relate to text findings;

2) SNOMED-CT Clinical ontology system, and its normalizations and mappings to parent ontological domains;

3) NDC (National Drug Code) system for prescriptions and

4) Ontologies for procedures, such as the ICD-10 Procedure Coding System (ICD-10-PCS) and CPT (Current Procedural Terminology).

The talk will close with a quick one-minute example on how to train a Scala Spark / PySpark NLP pipeline, by using a basic ontology to create an ML classifier for a simulated EHR (Electronic Health Record) dataset.

About the speaker

Marcelo Tournier

Data Scientist at Apixio

Marcelo is a tech-seasoned Physician + Programmer who loves to build models in TensorFlow, Spark, Scala, Python, Java & R, bringing data to life. He is a Product Manager, with 10+ years building technologies (Apps, IoT, Machine Learning).

Marcelo championed the design of data-driven approaches for healthcare, winning national awards of excellence and one patent. He is also an advisor of Data science startups in Brazil.



Sessions: October 5 – 7
Trainings: October 4, 12 – 15


Presented by