Communications of the ACM
Machine Learning Turns Up COVID Surprise
The analysis suggested that long-term respiratory failure and the risk of secondary infection are much more common in COVID-19 patients than are cytokine storms.
Credit: iStock
Researchers at Northwestern University and the University of Wisconsin used machine learning (ML) to analyze daily pneumonia cases in an intensive care unit (ICU) based on electronic health records, and gained surprising insights about COVID-19 patients.
By clustering patient days, the ML analysis suggests COVID-19 patients in the ICU experience long-term respiratory failure and the risk of secondary infection more frequently than cytokine storms, severe immune reactions in which the body releases too many cytokines (small proteins important in cell signaling) into the blood too quickly
The researchers developed the CarpeDiem ML framework using the Jupyter Notebook platform, compiling a dataset of 44 distinct clinical parameters for each patient day, while clustering yielded 14 cohorts with different signatures of six types of organ dysfunction.
Northwestern's Catherine Gao said the ML algorithms the researchers used helped them "see clear patterns emerge that made clinical sense."
From IEEE Spectrum
View Full Article
Abstracts Copyright © 2023 SmithBucklin, Washington, DC, USA
No entries found