JMIR Publications

ABSTRACT

Background:

There is significant heterogeneity in disease progression among hospitalized COVID-19 patients. The pathogenesis of SARS-CoV-2 infection is attributed to a complex interplay between virus and host immune response that in some patients unpredictably and rapidly leads to “hyperinflammation” associated with increased risk of mortality. The early identification of patients at risk of progression to hyperinflammation may help inform timely therapeutic decisions and lead to improved outcomes.

Objective:

The primary objective of this study was to use machine learning to reproducibly identify specific risk-stratifying clinical phenotypes across hospitalized COVID-19 patients and to compare treatment response characteristics and outcomes. A secondary objective was to derive a predictive classification model using early encounter data that may be useful in informing optimal COVID-19 bedside clinical management.

Methods:

This is a retrospective analysis of electronic health record (EHR) data of adult patients (N= 4379) who were admitted to a Johns Hopkins Health System Hospital for COVID-19 treatment in the 2020-2021 timeframe. The phenotypes were identified by clustering 38 routine clinical observations recorded during inpatient care. To examine the reproducibility/validity of the derived phenotypes, patient data were randomly divided into two cohorts, and k-means clustering analysis was performed independently for each cohort. A predictive phenotype classifier using the Gradient Boosting Machine (GBM) method was derived using observations recorded during the first 6 h following admission.

Results:

Two phenotypes (designated as P1 and P2) were identified in patients admitted for COVID-19 in both the training and validation cohorts with similar distributions of features and correlations with biomarkers, treatments, comorbidities, and outcomes. In both the training and validation cohorts, P2 patients were older, had elevated markers of inflammation, and were at an increased risk of requiring intensive care unit-level care, developing sepsis, and mortality, compared to P1 patients. The GBM phenotype predictive model yielded an area under the curve (AUC) of 0.89 and a positive predictive value (PPV) of 0.83.

Conclusions:

K-means clustering was effective in identifying two phenotypes with distinct treatments/interventions and outcomes in a large cohort of hospitalized COVID-19 patients. Additionally, a GBM machine learning classifier model using readily available early encounter data accurately assigned patients to phenotypes, suggesting that the application of these models in a clinical setting may provide valuable prognostic information that could inform personalized COVID-19 management. While future studies and trials are needed to validate the clinical utility of phenotype assignment, it would seem reasonable to implement successfully validated ML algorithms in extant EHR systems as a tool to support those trials. Clinical Trial: not applicable