AI Health Virtual Seminar: Augmenting traditional population-based screening with AI: use case of identifying health system patients with poor functional status

November 9, 2022
12:00 pm to 1:00 pm
Virtual

Event sponsored by:

AI Health
+DataScience (+DS)
Biomedical Engineering (BME)
Biostatistics and Bioinformatics
Center for Computational Thinking
Computer Science
Electrical and Computer Engineering (ECE)
Pratt School of Engineering

Contact:

Duke AI Health

Share

Speaker:

Juliessa M. Pavon, MD, MHS, Associate Professor of Medicine (Geriatrics) and Head & Neck Surgery and Communication Sciences; Senior Fellow in the Center for the Study of Aging and Human Development; Durham VA Health System GRECC with host
Mary Solomon, BS, MS, Duke AI Health Fellow

Artificial intelligence algorithms can complement traditional screening methods and result in a population-based strategy for identifying high risk patients within large health systems - including the use case of identifying patients with poor physical function and at risk for early palliative care needs.

In this 1-hour seminar, we'll discuss:
(1) using electronic health record (EHR) data for a data-driven approach to categorize patients with poor functional status,
(2) applying unsupervised machine learning algorithms for clustering visualization of the data,
(3) training and testing a supervised machine learning algorithm to distinguish between patient functional status types,
(4) learning features of EHR clinical data that predict functional status types, and
(5) evaluation of the model for potential health inequities.

Our session will benefit attendees by increasing their understanding of current applications of AI using health system EHR data, as well as the initials steps for piloting implementation of the model for population-based screening.