Natural Language Processing for Medical Billing Code Prediction

It is extremely challenging for clinicians and health researchers to extract insights from unstructured text
data at scale. Recently, there has been a significant advancement in Natural Language Processing. Large
deep-learning models, most famous of which are the transformers, generate pre-trained embeddings and
are developed to extract insights from massive amounts of text.
This project develops AI models to assist the billing code assignment tasks. We focus on the ability of the
pre-trained embeddings to assist the identification and appropriate assignment of billing codes for the
insurable services defined by the Ontario Health Insurance Plan (OHIP). Training a model to generate rich
enough embeddings for the automatic coding task has the potential to improve predictive models for other
clinical NLP tasks, such as symptom extraction, cohort analysis, disease tracking, adverse effect
identification.

Faculty Supervisor:

Helen Chen

Student:

Partner:

IntelAGENT Billing

Discipline:

Computer science

Sector:

Professional, scientific and technical services

University:

University of Waterloo

Program: