Fig. 5: Results of different models evaluated on 2019 individuals with confirmed RT-PCR results for diverse viruses, including Influenza-A, Influenza-B, Influenza-H1N1, and Seasonal Influenza.
From: Prediction of SARS-CoV-2-positivity from million-scale complete blood counts using machine learning
a Model trained only on SARS-Cov-2 data. CBC (−) includes only COVID-19 (−). b Model trained using data of diverse viruses, including SARS-Cov-2. CBC (−) also includes viruses other than SARS-Cov-2. c The stacked model. CBC (−) also includes viruses other than SARS-Cov-2. Specialized models are trained using half of the training sets, and then these specialized models are combined using the other half of the training sets. N = 11 116 CBCs.
