Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2023 Apr 29:2022:1257-1266.
eCollection 2022.

Learning Clinical Concepts for Predicting Risk of Progression to Severe COVID-19

Helen Zhou  1 , Cheng Cheng  1 , Kelly J Shields  2 , Gursimran Kochhar  3 , Tariq Cheema  3 , Zachary C Lipton  1 , Jeremy C Weiss  1
Affiliations

Learning Clinical Concepts for Predicting Risk of Progression to Severe COVID-19

Helen Zhou et al. AMIA Annu Symp Proc. .

Abstract

With COVID-19 now pervasive, identification of high-risk individuals is crucial. Using data from a major healthcare provider in Southwestern Pennsylvania, we develop survival models predicting severe COVID-19 progression. In this endeavor, we face a tradeoff between more accurate models relying on many features and less accurate models relying on a few features aligned with clinician intuition. Complicating matters, many EHR features tend to be under-coded degrading the accuracy of smaller models. In this study we develop two sets of high-performance risk scores: (i) an unconstrained model built from all available features; and (ii) a pipeline that learns a small set of clinical concepts before training a risk predictor. Learned concepts boost performance over the corresponding features (C-index 0.858 vs. 0.844) and demonstrate improvements over (i) when evaluated out-of-sample (subsequent time periods). Our models outperform previous works (C-index 0.844-0.872 vs. 0.598-0.810).

PubMed Disclaimer

Figures

Figure 1.
Figure 1.
Screenshot of interactive Sankey diagram showing how raw features (first column) translate into clinical concepts (second column), and how both are ultimately used in each model (third column). Magnitude of coefficients correspond to flow thickness, positive log HRs are blue, and negative log HRs are red. Black flows indicate positive anchors for the corresponding concept. Visit acmilab.org/severe_covid to interact with the full diagram.
Figure 2.
Figure 2.
Kaplan Meier survival curves for the high (top 10%), medium (top 10-25%), and low (bottom 75%) risk groups using predictions from the LC + All Features model (left) and the LC only model (right). Counts at the bottom show the number of individuals who are at risk, are censored, or experienced the severe COVID-19 event across time.
Figure 3.
Figure 3.
One-calibration of the LC + All Features model (left) and LC model (right) at 14 days, binned into ten groups. Red dotted line corresponds to perfect calibration.
Figure 4.
Figure 4.
D-calibration histogram of LC + All Features model (left) and LC model (right), binned into ten groups. In a completely D-calibrated model, all of the horizontal bars should be at 0.10.
See this image and copyright information in PMC

References

    1. Docherty AB, Harrison EM, Green CA, Hardwick HE, Pius R, Norman L, et al. Features of 20 133 UK patients in hospital with covid-19 using the ISARIC WHO Clinical Characterisation Protocol: prospective observational cohort study. BMJ. 2020 May 22;369:m1985. - PMC - PubMed
    1. Henry BM, de Oliveira MHS, Benoit S, Plebani M, Lippi G. Hematologic, biochemical and immune biomarker abnormalities associated with severe illness and mortality in coronavirus disease 2019 (COVID-19): a meta-analysis. Clin Chem Lab Med. 2020 Jun 25;58(7):1021–8. - PubMed
    1. King JT, Jr, Yoon JS, Rentsch CT, Tate JP, Park LS, Kidwai-Khan F, et al. Development and validation of a 30-day mortality index based on pre-existing medical administrative data from 13,323 COVID-19 patients: The Veterans Health Administration COVID-19 (VACO) Index. PLoS One. 2020 Nov 11;15(11):e0241825. - PMC - PubMed
    1. Xu W, Sun NN, Gao HN, Chen ZY, Yang Y, Ju B, et al. Risk factors analysis of COVID-19 patients with ARDS and prediction based on machine learning. Sci Rep. 2021 Feb 3;11(1):2933. - PMC - PubMed
    1. Zhou H, Cheng C, Lipton ZC, Chen GH, Weiss JC. 18th International Conference on Artificial Intelligence in Medicine (AIME) 2020, Proceedings. Springer-Verlag; 2020. Mortality Risk Score for Critically Ill Patients with Viral or Unspecified Pneumonia: Assisting Clinicians with COVID-19 ECMO Planning; pp. 336–47.

LinkOut - more resources