The clinical outcome of SARS-CoV-2 infection varies widely between individuals. Machine learning models can support decision making in healthcare by assessing fatality risk in patients that do not yet show severe signs of COVID-19. Most predictive models rely on static demographic features and clinical values obtained upon hospitalization. However, time-dependent biomarkers associated with COVID-19 severity, such as antibody titers, can substantially contribute to the development of more accurate outcome models. Here we show that models trained on immune biomarkers, longitudinally monitored throughout hospitalization, predicted mortality and were more accurate than models based on demographic and clinical data upon hospital admission. Our best-performing predictive models were based on the temporal analysis of anti-SARS-CoV-2 Spike IgG titers, white blood cell (WBC), neutrophil and lymphocyte counts. These biomarkers, together with C-reactive protein and blood urea nitrogen levels, were found to correlate with severity of disease and mortality in a time-dependent manner. Shapley additive explanations of our model revealed the higher predictive value of day post-symptom onset (PSO) as hospitalization progresses and showed how immune biomarkers contribute to predict mortality. In sum, we demonstrate that the kinetics of immune biomarkers can inform clinical models to serve as a powerful monitoring tool for predicting fatality risk in hospitalized COVID-19 patients, underscoring the importance of contextualizing clinical parameters according to their time post-symptom onset. Author summary SARS-CoV-2 infected patients present with diverse clinical profiles, ranging from asymptomatic to severe respiratory failure and death. Early detection of high-risk patients is fundamental to tailor therapeutic interventions that anticipate disease progression and prevent poor outcomes. Machine learning can assist health workers in triaging patients by bringing together multiple factors, describing the patient’s health, into a single model capable of predicting the most likely outcome. This can be particularly relevant in surge settings where clinical resources must be efficiently utilized. To date, most models predict COVID-19 outcomes using patient data obtained upon hospital admission. However, clinical data obtained longitudinally during hospitalization can provide a wealth of information to build more precise models. With this in mind, we monitored disease progression in 147 COVID-19 patients during hospitalization by frequently collecting clinical parameters. We show that models trained on longitudinally monitored immune biomarkers predicted mortality and were more accurate than models based on demographic and clinical data obtained upon hospital admission. Our work encourages the development of a broader computational framework that combines patient clinical data from hospital admission with longitudinally monitored biomarkers collected throughout the hospitalization to better assist health care workers with daily prognostication of COVID-19 patients.
【초록키워드】 COVID-19, Biomarker, Biomarkers, Health care, Respiratory failure, Mortality, health workers, Hospitalization, Decision making, SARS-COV-2 infection, severity, machine learning, health care workers, severe respiratory failure, neutrophil, COVID-19 severity, C-reactive protein, risk, outcome, anti-SARS-CoV-2, immune, Predictive value, Clinical outcome, outcomes, Disease progression, lymphocyte, Kinetics, Health, Predictive model, Asymptomatic, White blood cell, healthcare, Early detection, Patient, Factors, death, Health care worker, Hospital admission, WBC, Severity of disease, urea nitrogen, information, resource, parameters, predict, COVID-19 patients, Antibody titers, High-risk patients, fatality, Clinical parameters, health worker, PSO, COVID-19 patient, machine learning models, blood cell, Predictive, nitrogen, urea, Support, patient data, Blood urea nitrogen, Temporal analysis, hospitalized COVID-19 patients, Clinical data, SARS-CoV-2 infected patients, post-symptom onset, profiles, therapeutic interventions, therapeutic intervention, Demographic and clinical data, lymphocyte counts, machine, mind, MOST, clinical parameter, Spike IgG titers, high-risk patient, Prevent, feature, predicted, collected, contribute, individuals, assist, build, SARS-CoV-2 infected patient, with COVID-19, 【제목키워드】 COVID-19, Biomarker, outcome, immune, predict, the timing,