Machine Learning Prediction for Supplemental Oxygen Requirement in Patients with COVID-19

Introduction: The coronavirus disease (COVID-19) poses an urgent threat to global public health and is characterized by rapid progression even in mild cases. In this study, we investigated whether machine learning can be used predict which patients will have a deteriorated condition require oxygenation asymptomatic or cases of COVID-19. Method: This single-center, retrospective, observational study included COVID-19 admitted the hospital from February 1, 2020, May 31, who were either presented with symptoms did not oxygen support on admission. Data patient characteristics vital signs collected upon We seven algorithms, assessed their capability exacerbation, analyzed important influencing features using best algorithm. Results: total, 210 study. Among them, 43 (19%) required therapy. Of all models, logistic regression model had highest accuracy precision. Logistic analysis showed that 0.900, precision 0.893, recall 0.605. most parameter for predictive was SpO2, followed age, respiratory rate, systolic blood pressure. Conclusion: developed as triage tool clinicians detect high-risk earlier. Prospective validation studies are needed verify application clinical practice.