acm-header
Sign In

Communications of the ACM

ACM TechNews

Researchers Analyze 53 Million Points of Clinical Data with Embedding Technique


View as: Print Mobile App Share:
doctor at computer interface surrounded by medical data icons, illustration

Researchers say their technique surpasses current computational models.

Researchers at Children's Hospital of Philadelphia (CHOP) and Drexel University used a phenotype embedding technique to analyze 53 million patient notes from more than 1.5 million individuals to identify similar medical histories.

The researchers employed CHOP's Arcus clinical research platform, designed to show how sets of data overlap at a massive scale. The published analysis reveals 9,477 distinct phenotypes, demonstrating a high level of correspondence with various experts' evaluation.

The experiment converted complex phenotypes from the Human Phenotype Ontology format into arrays of clinical data, yielding a model that will facilitate efficient simulation for later downstream tasks requiring deep phenotyping.

The algorithm developed for the analysis "will allow us to use machine learning in tandem with existing methods to analyze risks and patient prognoses in a more efficient manner at large scale," says Dr. Ingo Helbig, a pediatric neurologist at CHOP.

From Children's Hospital of Philadelphia
View Full Article

 

Abstracts Copyright © 2023 SmithBucklin, Washington, D.C., USA


 

No entries found

Sign In for Full Access
» Forgot Password? » Create an ACM Web Account