event-icon

Oral Presentations

Ontologies Enabling Computable Tables

8:30 AM–10:00 AM Nov 19, 2019 (US - Eastern)

Gunston

Description

Machine learning algorithms require as input curated “features” that summarize and group the source data. Developing such a computable table is predicated on a deep understanding of coding systems (both local and standardized), value sets, and local knowledge. Here we develop and pilot tools for Informatics for Integrating Biology and the Bedside (i2b2) that leverage the ontology system to build such tables: a computable-table builder, a bulk export tool, and an ontology-metadata explorer.

Learning Objective: After participating in this activity, the participant will be better able to leverage their institution’s data resources to develop ontology-driven computable tables for input to machine learning algorithms.

Authors:

Jeffrey Klann (Presenter)
Harvard Medical School

Nich Wattanasin, Partners Healthcare
Michael Mendis, Partners Healthcare
Matthew Joss, Partners Healthcare
Hossein Estiri, Harvard Medical School
Kavishwar Wagholikar, Harvard Medical School
Shawn Murphy, Partners Healthcare

Presentation Materials:

Tags