Value Proposition:
· High accuracy in predicting both all-cause mortality/cardiac arrest and thromboembolic events in hospitalized COVID-19 patients, with clinically meaningful early warning times.
· Real-time patient risk assessment based on clinical data, via a continuously updating risk score for an outcome within a specific timeframe.
· Easily interpretable output with identified patient-specific risk drivers, unlike many “black box” machine learning tools.
· Proven generalizability, with strong performance across patient data from five different hospitals.
Technology Description
· Researchers at Johns Hopkins have developed COVID-HEART, a machine learning-based risk predictor that forecasts adverse cardiovascular events in hospitalized COVID-19 patients, where timely intervention is critical but current tools are inadequate. COVID-HEART uses clinical data such as vital signs, electrocardiograms, and comorbidities to continuously update patient risk scores, enabling early evidence-based intervention based on subtle changes in biomarkers during the disease process. The model highlights key features driving the risk score, increasing interpretability so clinicians can target underlying causes. The data demonstrates that COVID-HEART can accurately predict all-cause mortality/cardiac arrest and thromboembolic events within a clinically useful timeframe.
Unmet Need
· Hospitalized COVID-19 patients are at high risk for serious cardiovascular complications such as cardiac arrest and thromboembolic events, yet clinicians currently lack tools to identify these events before they occur. Other predictive models fail to account for the dynamic course of the disease, do not focus on cardiovascular events, update periodically rather than in real-time, and offer limited guidance for timely, targeted intervention. Therefore, there is a strong need for a real-time risk prediction model to be developed to address the challenge of identifying hospitalized COVID-19 patients at risk of adverse cardiovascular events in time for clinical intervention.
Stage of Development
· The technology has been developed and validated.
Data Availability
· Data available upon request
Publication
Shade et al., Real-Time Prediction of Mortality, Cardiac Arrest, and Thromboembolic Complications in Hospitalized Patients With COVID-19, JACC: Advances, June 2022
