Construct validity of five sentiment analysis methods in the text of encounter notes of patients with critical illness release_s3uorbjqjrbmvixdonkijudiqe

by Gary Weissman, Lyle Ungar, Michael Harhay, Katherine Courtright, Scott Halpern

Released as a post by Cold Spring Harbor Laboratory.

2018  

Abstract

Sentiment analysis may offer insights into patient outcomes through the subjective expressions made by clinicians in the text of encounter notes. We analyzed the predictive, concurrent, convergent, and content validity of five sentiment methods in a sample of 791,216 multidisciplinary clinical notes among 40,602 hospitalizations associated with an intensive care unit stay. None of these approaches improved early prediction of in-hospital mortality. However, positive sentiment measured by Pattern (OR 0.09, 95% CI 0.04 - 0.17), sentimentr (OR 0.37, 95% CI 0.25 - 0.63), and Opinion (OR 0.25, 95% CI 0.07 - 0.89) were inversely associated with death on the concurrent day after adjustment for demographic characteristics and illness severity. Median daily lexical coverage ranged from 5.2% to 20.5%. While sentiment between all methods was positively correlated, their agreement was weak. Sentiment analysis holds promise for clinical applications, but will require a novel domain-specific method applicable to clinical text.
In application/xml+jats format

Archived Files and Locations

application/pdf  270.4 kB
file_g6x4uvbaojcb7nmcylabjiohu4
web.archive.org (webarchive)
www.biorxiv.org (web)
application/pdf  269.4 kB
file_fvwdvqezqvfclag3ehortcrc7e
www.biorxiv.org (web)
web.archive.org (webarchive)
Read Archived PDF
Archived
Type  post
Stage   unknown
Date   2018-04-27
Work Entity
access all versions, variants, and formats of this works (eg, pre-prints)
Catalog Record
Revision: 100eff02-03a9-4690-a242-83b594bc1da3
API URL: JSON